RESERVOIR FLOOD RISK INDEX:

Transcription

1 WLE MEKONG Impact of hydropower on Mekong floods & droughts (MK12) RESERVOIR FLOOD RISK INDEX: ASSESSING THE FLOOD RISK AND CAPACITYFOR CONTROL INMEKONG HYDRO- ELECTRICRESERVOIRS MK12 TECHNICAL NOTE 1 Tarek Ketelsen, Arun Parameswaran, Simon Tilleard, Alex Kenny, Mai Ky Vinh, Nguyen Mong Thuy, Luke Taylor April, 2014 i

2 WLE MEKONG Impact of hydropower on Mekong floods & droughts (MK12) Authors Produced by Produced for Suggested citation More information Image Project Team Copyright Acknowledgements Tarek Ketelsen, Arun Parameswaran, Simon Tilleard, Alex Kenny, Mai Ky Vinh, Nguyen Mong Thuy, Luke Taylor ICEM International Centre for Environmental Management Mekong Challenge Program for Water & Food Project 12: The impact of hydropower on floods and droughts in the Mekong Region Ketelsen, T., Parameswaran, A., Tilleard, S.,Kenny, A.,Vinh, M.K, Thuy, N.M., Taylor L Reservoir Flood Risk Index - a tool for assessing the flood risk and capacity for control in Mekong hydro-electric reservoirs. Technical note: Challenge Program on Water & Food Mekong project MK12 The impact of hydropower on floods and droughts in the Mekong Region. ICEM International Centre for Environmental Management, Hanoi Vietnam, Google Earth, T. Ketelsen, P.J. Meynell ICEM Tarek Ketelsen, Arun Parameswaran, Simon Tilleard, Alex Kenny, Mai Ky Vinh, Nguyen Mong Thuy, Paul Wyrwoll, Luke Taylor IWRP Lê Thanh Hà, Ha Thanh Lan EDI Piseth Vann, Prom Tola, Lay Reth 2014 ICEM - International Centre for Environmental Management 6A Lane 49, Tô Ngoc Vân Tay Ho, HA NOI Socialist Republic of Viet Nam Long term daily flow data in Mekong tributaries catchment and technical specifications on Mekong hydropower projects was obtained with gratitude from the Mekong River Commission. We would also like to acknowledge the cooperation of Nikki West from the CPWF Mekong Naga House for information on reservoir locations and parameters and PECC1 for provision of some additional gauging station data for the Central Highland area. Funding for this project was from the Mekong river component of the CGIAR Challenge Program on Water and Food. ii

3 WLE MEKONG Impact of hydropower on Mekong floods & droughts (MK12) Executive summary Flooding is both a blessing and a curse to the ecosystems and communities of the Lower Mekong Basin (LMB).The Mekong River has one of the most varied hydrological regimes of the world s major river systems. This variability is driven by the combination of two monsoon regimes and the westpacific storm system, which induces a dual signal in Mekong flood patterns the monsoon which is predictable and occurs on an annual basis, and storm events which are infrequent and significantly larger in magnitude. Flooding associated with extreme storm events is a serious threat to the lives and livelihoods of Mekong communities.annual seasonal flooding has made the Mekong basin one of the most biodiverse and productive river basins on the planet. However, during extreme years, tropical storms and cyclones originating in the Pacific Ocean can induce short duration precipitation events that are much greater than average peak daily rainfall. Reservoirs have been utilised for centuries in the Mekong Basin to regulate seasonal water availability and provide water for consistent, year-round human use. For agriculture, water supply and more recently hydroelectricity, the storage and distribution of wet season flows for use in the dry season has led to tangible and significant improvements in livelihoods, agricultural productivity and energy security since the time of the Khmer Kingdom. However, having been designed and managed for regular climate, the performance of these reservoirs in managing extremes in hydro-climate has been poorly understood. In theory, Water Supply Infrastructure (WSI) can regulate flood and drought events by changing the timing of downstream releases and reducing flood peaks.in practice, WSI in the Mekong can often exacerbate extreme floods and droughts due to optimisation for single use options (i.e. electricity production) or uncoordinated management and in some cases conflict between upstream and downstream water users. Existing reservoir management highlights the conflicting risks of securing electricity production and regulating flood peaks and the urgent need for improved incorporation of flood control into the design and management of Mekong Reservoirs. The highly variable rainfall conditions of the monsoon pressures operators to restock the reservoir as soon as possible in the wet season, as levels less than full supply level (FSL) at the end of the wet season would result in a reduction to year-round electricity production. At the same time, the early stocking of the reservoir to FSL burdens the dam with a greater risk of insufficient capacity to cope with a major storm event. When a cyclone or major storm impacts the area late in the season, the excess water is released immediately as there is no available storage, which can cause significant flooding issues. There is also the potential for worsened downstream flood conditions if hydraulic infrastructure underperforms or if human judgement on the timing for initiating releases is slow. The rapid rate of hydropower development further compounds the problem of integrating flood control into reservoir design and operation. With at least 130 large projects planned for the LMB a response by governments to incorporate flood control into the design and management of large reservoirs is hampered by the scale required. Where should Government regulators start in retrofitting and re-managing existing reservoirs and which planned reservoirs have the greatest potential to harness flood control or in which areas of the basin are flood threats greatest and control measures most urgently needed? iii

4 WLE MEKONG Impact of hydropower on Mekong floods & droughts (MK12) THE RESERVOIR FLOOD RISK INDEX A SCREENING TOOL FOR SETTING PRIORITIES TO ENHANCE FLOOD CONTROL IN LARGE MEKONG RESERVOIRS This study develops a Reservoir Flood Risk Index (the Index) to help stakeholders set priorities in terms of strategic flood control response for hydropower in the LMB. The Reservoir Flood Risk Index is a rapid, basin-wide screening tool which allows developers, government and communities to understand: (i) (ii) (iii) Which reservoirs have the capacity to control floods, Which reservoirs are in high flood yield catchments, and Which reservoirs have the highest potential for downstream damage? It also identifies which existing or proposed reservoirs have the greatest potential to be used for multiple use flood control. Of the 136 projects included in the MRC database a subset of 67 was selected for inclusion in the assessment based on the availability of reliable information on design specifications needed to assess Flood Control, Natural Flood Threat and Flood Impact Potential. The Index aggregates three sub- Indices for each of the large hydropower reservoirs comprised of nine indicators derived from publicly available data: (i) (ii) (iii) Flood Control: what is the structural and technical capacity of the WSI to control flood events? This relates to indicators such as the reservoir emergency storage, regulating and discharge capacity. Natural Flood Threat: what is the nature and magnitude of flood events in the subcatchments of the Mekong? This is generally measured in terms of flood frequencies and arrival times and is related to parameters such as the catchment slope, vegetation types and precipitation dynamics. Flood Impact Potential: what are the land use and human habitation characteristics of the at-risk areas downstream of the WSI? This relates to indicators such as population, poverty, agricultural land area and ecological priorities. Figure A: Indicators utilised in the Reservoir Flood Risk Index: (RED) Reservoir design and infrastructure characteristics which influence a reservoir s capacity to store flood waters, (GREEN) catchment characteristics that measure timing and magnitude of incoming flood events, (ORANGE) land cover and socio-economic characteristics of a 5km riparian corridor within 50km downstream of the dam structure 1. Probable Maximum Flood (PMF) inflow yr ARI inflow 3. Catchment time of concentration 4. Reservoir emergency storage 5. Live storage vs mean annual flow 6. Discharge capacity vs flood inflow 7. Downstream riparian Population (and poverty levels) 8. Downstream riparian agricultural land 9. Downstream ecosystem services iv

5 WLE MEKONG Impact of hydropower on Mekong floods & droughts (MK12) Scores for each of the 9 indicatorsare standardised into a rank 0-1 for each of the three sub-indices: (i) Natural Flood Threat, (ii) Flood Controland (iii) Flood Impact potential. The Natural Flood Threat and Flood Control scores are used to characterise the Downstream Flood Potential, which is then combined with the Flood Impact Potential to obtain the overall Reservoir Flood Risk Index score (Figure B). Figure B: Aggregation of thereservoir Flood Risk Index: Information on 67 reservoirs and their catchments is used to characterise flood threat, control potential and impact potential, which are then aggregated to determine downstream flood potential and the Reservoir Flood Index. The Reservoir Flood Risk Index has been developed to provide opportunities for flood management to be considered at various points in the project design cycle. At the first stage of the development process, the Index can act as a screening tool for governments to set priorities on where and how flood control should be integrated into the project development cycle to maximise the benefit. In addition, the Index can be used during the environmental review phase for designers to prioritise flood control in reservoirs where it is most needed. Last, the Index will allow hydropower developers to estimate the flood risk in a sub-catchment at the pre-feasibility and scoping phases of project development before firm agreements on electricity production and designs are finalised so that the importance/need for flood control is considered. This paper is supported by a number of products available from including: (i) an interactive map which summarises data for each of the 67 projects, (ii) an excel-based Mekong hydropower Reservoir Flood Risk Index database, (iii) supporting annexes detailing hydrological calculations made for gauged and ungauged catchments in the LMB, and (iv) case study reports from two WSI reservoirs in Cambodia and Viet Nam. MAIN FINDINGS FOR THE RESERVOIR FLOOD RISK INDEX The 67 dams included in the study can be geographically grouped into three regions: (i) the 3S region of the Se San, Se Kong and Sre Pok river basins and including the small southern Lao catchments adjacent to the Se Kong; (ii) the Northern Lao catchments joining the Mekong River upstream of Vientiane/Nam Ngum (Nam Lik, Nam Khan, Nam Ngum, Nam Suang etc); and, (iii) the Lao catchments of the northern Annamite ranges (Nam Ngiep to Nam Theun). Table A summarises the top and bottom performing reservoirs for each of:natural flood threat, flood control potential and downstream impact potential. Catchment NaturalFlood Threat Catchment Natural Flood Threat is highest in the 3S region of Lao PDR, Cambodia and Viet Nam especially the Se Kong and surrounding small catchments of southern Lao PDR while flood threats are generally lower in the Northern Annamites and Northern Lao PDR. The study found that some 40% of the dams assessed have design peak inflows lower than the 100year ARI flow. Thesedams are at higher risk of aggravating downstream flooding during extreme rainfall events. In addition, a v

6 WLE MEKONG Impact of hydropower on Mekong floods & droughts (MK12) quarter of the dams assessed were built in catchments with quick response times, which means operator response time available to manage flood flows is also short (in the order of a few hours). TableA: Summary rankings for Flood threat, Control and Impact potentials. A B C TOP TEN Catchment NaturalFlood Threat Reservoir Flood Control potential Downstream Flood Impact Potential 1 Lower Se San2 Xe Bang Nouang Xe Labam 2 Xe Kong 3d Nam Hinboun 2 Lower Sesan 2 3 Xe La Bam Nam Poun Nam Khan 1 4 Xe Kong 3up Nam Ngum 1 Plei Krong 5 Xe Kong 4 Nam Mang 1 Xe Pon 3 6 Xe Don 2 Nam Mang 3 Xe Don 2 7 Xe Kong 5 Nam San 2 Nam Ngum 1 8 Xe Kaman 1 Nam Theun 4 XeSet 1 9 Xe Xou Buon Kuop Buon Tua Srah 10 Xe Kaman 2A Nam Theun 2 Nam Kong 2 BOTTOM TEN Catchment NaturalFlood Threat Reservoir Flood Control potential Downstream Flood Impact Potential 58 Nam Pouy Xe Kaman 2A Nam Mouan 59 Nam Khan 3 Xelabam Xe Kaman 2B 60 Nam Poun Dray Hlinh1 Nam Mang 3 61 Nam Ngum 4A Dray Hlinh 2 Se San 3 62 Nam Ngum 2 Xe Katam Nam Pot 63 Buon Kuop Nam Kong 2 Nam Leuk 64 Nam Suang 1 Nam Pay Xe Katam 65 Nam Khan 2 Nam Phak Nam Chian 66 Nam Ngum 1 Xeset 1 Nam Theun 2 67 Nam Khan 1 O Chum 2 Nam Ngiep 1 Reservoir Flood Control potential Reservoirs with the lowest potential to control floods also tend to be in the 3S and surrounding region especially the Se San and Lao components of the Se Kong and surrounding catchment. Specifically, the study found: 1. In the LMB some of the largest dams have some of the smallest relative capacity to store extreme flood flows in emergency situations. The dams at particular risk are those with low capacity to store water and a relatively large volume of live storage (water stored between FSL and LSL). In the event of dam failure, resultant inundation from this live storage volume would cause damage and potential loss of life to communities. Dak E Mule and Nam Ngum 5 Dams are both examples of high dams (>100 m) with large live storages ( mcm) but comparably very small capacity to store emergency flood water. These disproportionately large live storages could, under human error or mismanagement, result in downstream flood flows greater than natural conditions would create. 2. More than half of the dams assessed have a regulating capacity of less than 10%. In theory this can be compensated for through the provision of buffer storage, however, given that management practices in the LMB rarely account for flood buffer storage, the risk becomes more significant % of the 67 dams have a spillway discharge capacity that is greater than the 100 year ARI flood inflow, and 1% has a spillway discharge that is equal. These dams have a relatively lower risk. vi

7 WLE MEKONG Impact of hydropower on Mekong floods & droughts (MK12) Downstream Flood Potential The high flood risks and low control capacity of the 3S region mean that the greatest potential for dangerous downstream flooding falls within these areas of Southern Lao PDR, Eastern Cambodia and the central highlands of Viet Nam, specifically: 1. Dams located in medium, high and very high Natural Flood Threat areas of the Mekong Basin are generally not well designed for flood control. This is an important finding as it indicates that the first change needed to improve reservoir flood control in the region is improving the structural design of reservoirs. 2. The planned and existing reservoirs in the 3S basins have the highest Downstream Flood Potential, largely due to the combination of poor design capacity for reservoirs to accommodate flood inflows and the large magnitude of flood events in the 3S catchments. 3. Lower Sesan 2 + Lower Srepok 2 (Lower Sesan 2) Dam and Xe Kong 3d exhibit the highest Downstream Flood Potential of all LMB hydropower projects. 4. Three reservoirs in the Xe Kong cascade (3d, 3up and 5) were identified as very high and high risk, due to a lower flood control capacity and higher catchment flows. Five reservoirs in the Xe Kaman cascade (2A, 2B, 3, 4A and 4B) were found to be of high flood risk primarily due to their poor flood control (small live storage and emergency storage). 5. The Downstream Flood Potential for LMB reservoirs is exacerbated by poor capacity in existing hydropower designs to be utilised for flood control. For the majority of reservoirs assessed, most have low or very low capacity to regulate flood inflows. This means the majority of reservoirs are limited in their capacity to provide flood control for downstream areas because the incorporation of flood control as part of reservoir operations will require a redesign of reservoir and spillway infrastructure. Of the five reservoirs ranked as low risk, all have medium to high flood control and are located in small catchments with low flows. Downstream Flood ImpactPotential The highest flood impact potential is closely correlated with population distributions and therefore highest for reservoirs within Viet Nam (i.e. Se San and Sre Pok) and reservoirs closer to tributary confluences with the Mekong River (Lower Sesan 2, Xe La bam, Nam Khan 1, Nam Ngum 1), specifically. 1. The highest ranked reservoirs score high across all three indicators population, agricultural land, and ecological assets.lower Sesan 2, Xelabam and Nam Khan 1 were ranked with the highest downstream flood impact potential and scored high on all three indicators. 2. Irrespective of population a number of reservoirs in the 3S region have high downstream flood impact potentials because of very high scores for the ecological indicator. Plei Krong, Xe Pon 3, Nam Kong 2 and Upper Kontum are examples of dams in this region with high flood impact scores primarily due to high ecological indicator scores and moderate to high agricultural scores. 3. Some areas of extensive agricultural lands scored high based primarily on the agricultural indicator, for example, reservoirs in the Nam Ngum, upper Sre Pok and Xeset catchments. Reservoir flood index The Downstream Flood Potential and Flood Impact Potential results were combined to create a Reservoir Flood Risk Index with results summarised in Table B. Less than 5% of the 67 reservoirs assessed had high or very high scores for the reservoir flood index. There was one dam ranked as very vii

8 WLE MEKONG Impact of hydropower on Mekong floods & droughts (MK12) high flood risk (Lower Sesan 2) and two with high flood risk (Nam Khan 1 and Xe Labam). The majority of reservoirs (58%) have medium scores for the index, while the remaining 37% scored low. USING THE RESERVOIR FLOOD INDEX TO PRIORITISE RESPONSE The 67 reservoirs assessed in this study are being developed primarily for electricity production. The incorporation of a flood control use will impact on electricity production and consequently such initiatives should focus on reservoirs where flood control is most needed and likely to be least costly. Based on the results for the Reservoir Flood RiskIndex the study team have grouped the reservoirs into 5 categories of response, setting priorities for flood management based on the likelihood and consequence of flood (Table B): Category A: Category B1: Category B2: Category B3: Category C: Dams with high/very high Downstream Flood Potential and high/very high Flood Impact Potential should be singled out for prioritised intervention. These included the three dams Lower Sesan 2, Nam Khan 1 and Xelabam, all three of which are within the design phase and so have a high potential for redesign. Dams with high/very high Flood Impact Potential are a priority for action due to likely high impact on downstream populations if a flood occurs. These include the reservoirs of southern Lao PDR, the Se San River and Nam Ngum. Dams with high/very high Downstream Flood Potential are a priority for action due to the higher likelihood of floods occurring. These include more extensively the reservoirs of the 3S region. Dams with high Downstream Flood Potential but only medium Flood Impact Potential are a priority for action because it is likely that downstream floods will occur and the impacts may be relatively major. These include the remaining reservoirs of the 3S as well as the Nam Lik and Nam Ngum catchments in northern Lao PDR. Dams in Group C are a low priority because downstream floods are less likely to occur and if they do there is less potential for impacts downstream. Generally, most of the reservoirs in northern Lao PDR and northern Annamites fall within this category, with the exception of the Nam Khan 1, Nam Ngum 1 &2, and Nam Lik 1&2. viii

9 WLE MEKONG Impact of hydropower on Mekong floods & droughts (MK12) Table B: Findings for the Reservoir Flood Index and grouping of reservoirs based on the priorities for response Downstream flood potential Reservoir Flood Risk PRIORITIES FOR ACTION VERY HIGH HIGH MEDIUM LOW D = f (A,B) Lower Sesan 2 Xe Kong 3D Lower Sesan 2 A Dams with high/very high Downstream Flood Potential and high/very high Flood Impact Potential Xe Labam Houay Lamphan Xe Kaman 4B Nam Khan 1 Xelabam should be singled out for prioritised intervention Xe Kong 3up Dak E Mule Nam Phak Xe Kong 5 Xe Kaman 4A Nam Kong 2 Xe Kaman 2A O Chum 2 Xe Kaman 3 Xe Kaman 2B Xe Katam Xe Kong 4 Dray Hinh 1 Xe Pon 3 Xe Don 2 Xe Pon 3 Nam Ngum 1 B1 Dams with high/very high Flood Impact Potential Xe Don 2 Dray Hinh 2 Nam Ngiep 1 Plei Krong Xe Kong 3D are a priority for action due to likely high impact on downstream populations if a flood occurs Xe Kaman 1 Xe Set 3 Nam Leuk Nam Kong 2 Xe Kaman 2A O Chum 2 B2 Dams with high/very high Downstream Flood Xe Xou Nam Ngiep Nam Hinboun 1 Xe Kong 3up Xe Kaman 2B Xe Kaman 4B Potential are a priority for action due to the higher regulating dam likelihood of floods occurring Se San 3 Dam San 3B Nam San 2 Houay Lamphan Dak E Mule Xe Kaman 3 Xe Lanong 1 Nam Pouy Nam Theun 4 Nam Phak Xe Kaman 4A Xe Katam Xe Lanong 2 Nam Khan 3 Nam San 3 Xe Kong 5 Nam Lik 1 Nam Ngum 4A Nam Theun 2 Se San 4A Xe Xou Nam Son 3B B3 Dams with high Downstream Flood Potential but Nam Lik 2 Se San 3A Nam Mang 1 Upper Kontum Se San 3A Nam Hinboun 1 only medium Flood Impact Potential are a priority Upper Kon Tum Sesan 4A Nam Mang 3 Xe Set 1 Xe Lanong 1 Nam Ngum 2 for action because it is likely that downstream Xe Set 1 Nam Mouan Nam Suang 1 Buon Tua Srah Sre Pok 4 Nam Lik 2 floods will occur and the impacts may be relatively Nam pay Nam Ngum 5 Nam Khan 1 Buon Kuop Dray Hinh 1 Nam Lik 1 major Nam Pot Nam Suang 2 Nam Ngum 2 Xe Kong 4 Xe Set 2 Xe Lanong 2 Nam Chian Buon Tua Srah Nam Khan 2 Xe Kaman 1 Nam Mang 1 Sre Pok 4 Plei Krong Sre Pok 3 Xe Set 2 Buon Kuop Nam Poun Xe Bang Nouan Nam Hinboun 2 Nam Chian Nam Mang 3 C Dams in Group C are a low priority because Nam Ngum 1 Nam Hinboun 2 Bang Nouan Nam Son 3 Nam Pouy downstream floods are less likely to occur and if Se San 3 Nam Theun 2 Nam Khan 3 they do there is less potential for impacts downstream Nam Son 2 Dray Hinh 2 Nam Ngum 4A Nam Mouan Sre Pok 3 Nam Suang 1 Nam Pay Xe Set 3 Nam Khan 2 Nam Theun 4 Nam Ngiep 1 Nam Poun Nam Pot Nam Leuk Nam Suang 2 Nam Ngiep regulating dam ix

10 WLE MEKONG Impact of hydropower on Mekong floods & droughts (MK12) PRIORITY RESPONSES FOR ENHANCING FLOOD CONTROL USE OF LMB HYDROPOWER The reservoir flood index priorities identified above were developed by aggregating ranked scores for each reservoir s NaturalFlood Risk, Flood Control and Downstream Impact Potential. In general, the Index identified the reservoirs of the 3S basins and adjacent catchments of southern Lao together with the Nam Ngum and Nam Lik catchments as of highest need for the inclusion of flood control into reservoirs operations. Specifically, Lower Sesan 2, Nam Khan 1 and Xe Labam were identified as the highest priorities for response. The Downstream Flood Potential Index matrix helps to identity and prioritise management and design options to reduce the Downstream Flood Potential of a reservoir. The matrix can assist reservoir managers to understand if a dam needs structural design changes to improve flood control capacity or whether the design is sufficient and efforts should be focussed on improved management Figure C: Entry points for flood control into the Mekong hydropower planning cycle. Cycle restarts after Refurbishment/decommissioning, where there is an opportunity to Build, Operate, Transfer the infrastructure The opportunity to incorporate flood control into Mekong Reservoirs depends on the stage of project development. Figure C illustrates the major stages in a typical hydropower project cycle from planning through design, implementation, management and refurbishment/decommissioning. Based on this analysis, there are three main entry points for flood control: 1. Policy and regulation: should be implemented at the strategy, master plan and plan stages. This positively influences the flood control performance of all existing and planned dams 2. Design responses: should be input at the project plan and design steps. This is for planned dams where flood control can be incorporated into existing designs and intended management of the scheme. Dams with a high Natural Flood Threat and low Flood Control x

11 WLE MEKONG Impact of hydropower on Mekong floods & droughts (MK12) Potential can benefit from this. It is also relevant for existing dams that can be retrofitted or are being refurbished. 3. Management responses:should be included in the project implementation step. This would be for existing dams that are already operating or are being refurbished and transferred, have the potential capacity to reduce floods and are not already operated to do so. Dams with a high Natural Flood Threat and high Flood Control can benefit from this, as can dams with high Natural Flood Threat and low Flood Control although this would be to a lesser extent. Low Flood Control capacity reservoirs in high Natural Flood Threat catchments A larger number of dams are in high Natural Flood Threat areas but have limited capacity to control flood flows. There is significant potential to improve flood control through changes to design of the projects and their components. Design reform to include this should take place at the outset of project pre-feasibility assessments and concessions should be determined before firm agreements on the quantum of electricity production are negotiated so that these projects move forward as true multiple use dams. Inclusion of flood control design measures should focus on the reservoirs identified as medium to very high NaturalFlood Threat and very low Flood Control capacity. Of the dams categorised as Group B, the following have been found as priority based on the Reservoir Flood Risk Index results: First priority because of high Downstream Flood Potential and Flood Impact potential: Lower Sesan 2 and Xelabam Second priority because of: o o High Flood Impact potential: Xe Kong 3D High Downstream Flood Potential and very low to medium Flood Impact potential: Xe Kaman 2A, Xe Kaman 2B, Dak E Mule, Xe Kong 5, Xe Kaman 4A, O Chum 2, Xe Kaman 4B, Xe Kaman 3, Xe Kong 3up, Xe Katam, Nam Kong 2, Nam Phak and Houay Lamphan The planned projects in this list should be the focus of efforts to support developers to revise designs for enhanced flood control. Reservoirs in catchments with short response times Reservoirs with times of concentration of less than six hours mean that operators and downstream communities have limited time to respond once a large storm event begins. It is therefore important that these dams have good flood storage capacity to create a time buffer for flood preparations downstream and to begin controlled releases. Of the 67 dams assessed there were 11 reservoirs with low flood control capacity planned in catchments with short response times (Table C). Table C:Reservoirs with time of concentration less than six hours and low capacity to store and/or discharge flood flows TIME OF NAMES OF COMMISSIONED CONCENTRATION RESERVIOIRS NAMES OF PLANNED RESERVOIRS 0-2 hours N/A Nam Hinboun 2, Nam Pay and Nam Phak 2 4 hours O Chum 2 Nam San 3B, Dak E Mule, Xe Kaman 4A and Xe Kaman 4B 4 6 hours N/A Nam Chian, Xe Lanong 2, Xe Set 3 and Nam Mang 1 xi

12 WLE MEKONG Impact of hydropower on Mekong floods & droughts (MK12) These dams should be the focus of structural redesign and improved operational procedures to ensure better consideration of the quick response times of the catchments. Redesign and improved operational procedures may include: (i) Installation of gated spillways to allow improved control of discharge; (ii) Increasing the dam discharge capacity so that the dams can quickly release water on short notice; (iii) Creating a time buffer for downstream communities by specifically designing and operating for a flood storage volume; and (iv) Installing improved flood forecasting and early warning systems in parallel to construction of the reservoir. Reservoirs with high capacity for Flood Control in high Natural Flood Threat catchments Dams that have high potential for Flood Control and are located in catchments of high Natural Flood Threat may not currently include provisions for flood storage, but have the capacity for management reform to do so. From the analysis, it can be seen that none of the assessed dams in high flood risk catchments contained a high potential for the incorporation of flood control measures. For future as yet unplanned hydropower projects, potential management reforms to incorporate flood control could include: (i) Flood forecasting through hydrological monitoring to inform dam discharge operation. This could help to draw down the storage before a flood event. It should be noted that this is only an aid for decision-making. (ii) Early warning systems for downstream inhabitants to notify them on dam releases (whether the releases relate to normal operations or flood events). (iii) Installation of gated spillways to provide greater flexibility in controlling discharges, although gated spillways do not guarantee flood mitigation. (iv) Lowering of the dam design full supply level to incorporate dedicated flood buffer storage. This storage would remain empty unless it is required to store flood water. This would be discharged slowly to reduce the peak flow downstream. (v) Dam discharge rules that include consideration of timing of downstream tributary peak flows to avoid coinciding peak flows. Reservoirs with high capacity for flood control in moderate/low risk catchments Many dams in the basin could have high potential to manage downstream flooding with minor alterations in their operating regime. In addition, from a flood control point of view a number of existing dams (Nam Ngum 1 and Xe Bang Nouan) have a high potential within their existing design and would be well suited for use as a regional demonstration project in fully integrating flood control into reservoir operations and management. Nam Ngum 1dam in particular is part of a larger hydropower cascade which increases the complexity of management but also the downstream damage potential from poorly managed flood flows. xii

13 Table of Contents SECTION A: INTRODUCTION EXECUTIVE SUMMARY... III 1 INTRODUCTION Benefits and risks of flooding in the Lower Mekong Basin The role of hydro-electric reservoirs in downstream flooding The need for an integrated basin-level response to flood management THE RESERVOIR FLOOD RISK INDEX Management of hydro-electric reservoirs Purpose of the Reservoir Flood Risk Index Design of the index Overview Ranking indicator scores RESULTS OF THE INDEX Natural Flood Threat for hydropower contributing catchments Concept of Natural Flood Threat Overview of methodology for hydrological analysis in gauged catchments Indicator year ARI flow and ratio to peak design inflow Indicator 2 - Probable maximum flood Indicator 3 - Time of concentration Conclusions Flood Control Potential for hydropower reservoirs Indicator 4 - Reservoir emergency storage capacity Indicator 5 - Regulating capacity (live storage vs mean annual flow) Indicator 6 - Discharge capacity Conclusions Downstream Flood Potential: Combined Natural Flood Threat and Flood Control ranking Flood impact potential for hydropower reservoirs Use of the Zone of Influence Indicator 7 - Population Indicator 8 - Agriculture Indicator 9 Environment Reservoir Flood Risk Index: Combined Downstream Flood Potential and Flood Impact potential ranking BASIN WIDE PRIORITIES FOR INTEGRATING FLOOD CONTROL INTO MEKONG HYDROPOWER DEVELOPMENT Responding to the flood index Basin wide priority for action The Lower Sesan 2 + Lower Srepok 2 Dam Responding to the Downstream Flood Potential Index High control capacity dams in areas of high flood threat Group A Low control capacity dams in areas of high flood threat Group B Dams in areas of low flood threat Group C Dams in areas with short flood response times SUPPORTING NOTES ON INCORPORATING FLOOD CONTROL FOR MEKONG HYDROPOWER DAMS Understanding the limitations Methods of flood control Design measures Management measures xiii

14 5.2.3 Policy & regulation Entry points for change - the hydropower planning cycle REFERENCES ANNEX A: DAMS INCLUDED IN ASSESSMENT OF THE RESERVOIR FLOOD RISK INDEX ANNEX B: RESULTS OF THE FLOOD IMPACT POTENTIAL SUB-INDEX xiv

15 SECTION AINTRODUCTION 2

16 Tech note 1: Reservoir Flood Risk Index 1 INTRODUCTION 1.1 Benefits and risks of flooding in the Lower Mekong Basin Flooding is both a blessing and a curse to the ecosystems and communities of the LMB.The Mekong River has one of the most varied hydrological regimes of the world s major river systems (Welcomme, 1985). This variability is driven by the combination of two monsoon regimes and the west-pacific storm system, which induces a dual signal in Mekong flood patterns the monsoon which is predictable and occurs on an annual basis, and storm incidence which is infrequent and significantly larger in magnitude. Annual seasonal flooding has made the Mekong basin one of the most biodiverse and productive river basins on the planet.river flow varies annually by an order of magnitude between the dry and wet season resulting in a flood pulse hydrograph which sees the inundation ofhundreds of thousands of hectares in the Cambodian and Vietnamese floodplain every year (ICEM, 2013a). This seasonal cycling of the Mekong and Tonle Sap floodplain between terrestrial and aquatic states is one of the main drivers behind the system s high levels of biodiversity and productivity as flood waters facilitate a wide array of key ecosystem processes such as soil fertilisation, fish migrations, nutrient cycling and groundwater recharge. In extreme years, tropical storms and cyclones originating in the Pacific Ocean can induce short duration precipitation events that are much greater than average peak dailyrainfall. For example, tropical storm Wukong, which hit the Mekong Basin in 1996, saw daily wet season rainfall in low-mid elevation areas of southern Lao PDR in excess of 400mm compared to a long-term average daily peak rainfall in the order of 100mm (Figure 1). Between 1951 and 2009,in the order of100 storms have hit the LMB with approximately 40% passing through the central highlands region, and 40% through central and northern Lao PDR (Figure 2). Figure 1: n day rainfall observed at selected sites in LMB during Tropical Storm Wukong, Sept 1996 (Source: adapted from MRC, 2011) Flooding associated with extreme storm events is a serious threat to the lives and livelihoods of Mekong communities. The damage associated with flooding costs the LMB countries on average USD76 million per year with Cambodia and Thailand accounting for nearly 75% of this total (Figure 3). In proportion to GDP, Cambodia and Lao PDR are the most severely affected ofthe Mekong countries with flood damages costing approximately % of annual GDP compared to % in the 3

17 Tech note 1: Reservoir Flood Risk Index Mekong provinces of Thailand and Vietnam 1 (MRC, 2008).Under extreme conditions the costs can be an order of magnitude greater (Figure 3). Individual cyclones can cause hundreds of millions of dollars worth of damage and loss of human life. The year 2000 floods were estimated to have cost the Mekong countries more than USD 695 million (MRC, 2009). In addition, droughts and extreme floods caused by inter-annual climate variability have even been linked to the demise of civilisations such as the Great Angkor Kingdom of Cambodia (Brendan et al, 2010).In this case, an unusually long period of drought interspersed with intense flood events undermined the water supply and distribution systems of Western Cambodia, resulting in weakenedfood supplies that increased the Kingdom s susceptibility to political demise(brendan et al, 2010). Figure 2: Historic frequency and intensity of tropical storms and cyclones tracks for the LMB: (Source data: OCHA,2012) 25% 15% 41% 19% 1 For Thailand and Viet Nam GDP used in this figure was scaled to be proportionate of the national area within the LMB boundary. 4

18 Tech note 1: Reservoir Flood Risk Index Figure 3: Comparison of average and extreme flood damages for the LMB(Source: adapted from MRC, 2008) 1.2 The role of hydro-electric reservoirs in downstream flooding Reservoirs have been utilised for centuries in the Mekong Basin to regulate seasonal water availability and provide water for consistent, year-round human use. For agriculture, water supply and more recently hydroelectricity, the storage and distribution of wet season flows for use in the dry season has led to tangible and significant improvements in livelihoods, agricultural productivity and energy security. However, having been designed and managed for regular climate, the performance of these reservoirs in managing extremes in hydro-climate has been poor. Box 1: Impacts of hydro-electric reservoirs on flooding can be positive or negative depending on management & design the on-going case of Quang Nam in Vietnam illustrates the adverse downstream impacts of poor flood control from single-use hydro-electric projects. In theory, hydro-electric reservoirs have the potential to regulate seasonal flooding by storing wet season flows for release during the dry season. During regular flooding events this is often the case. However, during extreme events reservoirs can exacerbate downstream flooding.the pressure to store water during the wet season for maximized year-round electricity production leaves little capacity to manage inflows from storm events late in the season. In these circumstances, spillway releases can be comparable or in some cases larger than natural flood conditions.the result is damage to agriculture lands, property, infrastructure and loss of life. For example, during Storm Ketsana (2009) the A Vuong hydropower project in Quang Nam Vietnam rapidly released 150 million m³ which exacerbated downstream flooding, killing 163 people and resulting in property damages estimated at USD 786 million (thanh nhien news, 2013). Whether large releases are a function of how the reservoir is managed or limitations in the design of its hydraulic infrastructure is dam-specific, and remains poorly understood in the Mekong. This is one of the main areas of focus of the reservoir flood control index. 5

19 Tech note 1: Reservoir Flood Risk Index In the past 20 years, the hydrological regime of the Mekong Basin has been altered by the rapid expansion of WSI for hydro-electric generation and irrigation purposes. With the number of hydropower reservoirs and water supply abstraction points expected to quadruple by 2030; irrigation schemes expected to double their command area over the same time period (ICEM, 2010); and the frequency of droughts and floods expected to increase with climate change (ICEM, 2012a); understanding the implications of these structures on floods and droughts and implications for food security is an increasingly important priority. In theory, WSI can regulate flood and drought events by changing the timing of downstream releases and reducing flood peaks. In practice, WSI can often exacerbate extreme floods and droughts due to optimisation for single use options (i.e. electricity production) or uncoordinated management and in some cases conflict between upstream and downstream water users (Ward et al, in preparation; Box 1). 1.3 The need for an integrated basin-level response to flood management The rapid rate of hydropower development further compounds the problem of integrating flood control into reservoir design and operation. With at least 130 large projects planned for the LMB(Figure 5) the total number of small, medium and large irrigation and hydropower reservoirs in the LMB is in the order of 10,000 to 30,000. A response by governments to incorporate flood control into the design and management of reservoirs is hampered by the scale required. Where should Government regulators start in retrofitting and re-managing existing reservoirs and whichplanned reservoirs have the greatest potential to harness flood control or in which areas of the basin are flood threats greatest and control measures most urgently needed? Further, as a shared resource between Mekong countries, the implications of reservoir releases and flood control are transboundary. The rapid expansion of large hydropower on the Lancang River 2 since the mid-1990s has begun to transform the hydrological regime of the Mekong mainstream (Raesenan et al, 2010). Examples of transboundary connectivity are also emerging from the 3S river basins of the central highlands, where dam discharges for peaking operation resulted in rapid water level fluctuations in downstream countries, affecting riverine communities dependent on the use of and access to the riparian zone (GIZ, 2013). These developments demonstrate the high level of connectivity between national water resources and the need for bilateral or multilateral coordination between upstream and downstream flood management responses. This study develops a Reservoir Flood Risk Index (the Index) to help stakeholders answer these questions and set priorities in terms of strategic flood control response. The Reservoir Flood Risk Index is a basin-wide tool which allows developers, government and communities to understand the relative flood threat, control and impact potential associated with each of 67 large hydro-electric projects planned for the LMB (Figure 5). Of the 136 projects included in the MRC data base a subset of 67 were selected for inclusion in the assessment based on the availability of reliable information on design specifications needed to assess Flood Control, Natural Flood Threat and Flood Impact Potential. 2 The Chinese portion of the Mekong River is referred to as the Lancang River 6

20 Tech note 1: Reservoir Flood Risk Index Figure 4: Reservoir Flood Risk Index: the Index overlays basin-wide information on catchment hydrology and flood dynamics; design characteristics of 67 dams in the MRC hydropower data base; and downstream landuse and demographics to set priorities for integration of flood control into the development of the hydropower sector. 7

21 Figure 5:LMB large hydropower projects included in the Reservoir Flood Risk Index:(BLUE DOTS) existing, (PINK DOTS) proposed; (GREY DOTS) part of the MRC hydropower data base but not included in this study due to an incomplete data set. 8

22 Tech note 1: Reservoir Flood Risk Index 2 The Reservoir Flood Risk Index 2.1 Management of hydro-electric reservoirs Water Supply Infrastructure such as hydropower reservoirs have inherent multiple uses in applications such as water supply, agriculture and flood protection, however, in the LMB large reservoirs are typically developed for single source use of electricity generation. Typical reservoir management for electricity generation involves keeping the water levels in the storage as high as possible during the wet season so that there is sufficient water for electricity generation throughout the dry season. The key management stages are as follows (refer to numbering in Figure 6): Point 1: Dry Season Reservoir is full, or at Full Supply Level (FSL) - highest level during normal operations - due to the heavy rains in the preceding wet season. The dry season brings lower rainfall and plant operation begins to reduce thereservoir s storage level in order to sustain electricity production. Point 2: End of the Dry Season Storage level in reservoir continues to drop towards Low Supply Level (LSL). This is the lowest level in the reservoir, and also the point at which electricity can no longer be generated. Point 3: Start of the Wet Season High rainfall causes the storage level to rise and operators actively conserve inflows to re-stock the reservoir using the more predictable inputs from monsoon rains (June-August). Point 4:Start of the storm season Early in the wet season, the reservoir s FSL is reached and then maintained for use in the following dry season, leaving the reservoir with reduced capacity to manage additional inflows due to storms late in the wet season. Figure 6: Typical Reservoir Management for Electricity Generation(Image sources: Rasanen et al, 2013; VNCOLD,

23 Tech note 1: Reservoir Flood Risk Index Thisapproach to reservoir management highlights the conflicting risks of securing electricity production and regulating flood peaks. The highly variable rainfall conditions of the monsoon 3 pressures operators to restock the reservoir as soon as possible in the wet season, as levels less than FSL at the end of the wet season would result in a reduction to year-round electricity production. At the same time, the early stocking of the reservoir to FSL burdens the dam with a greater risk of insufficient capacity to cope with a major storm event. When a cyclone or major storm impacts the area late in the season, the excess water is released immediately as there is no available storage, which can cause significant flooding issues. There is also the potential for worsened downstream flood conditions if hydraulic infrastructure underperforms or if human judgement on the timing for initiating releases is slow (Ward et al, 2012). 2.2 Purpose of the Reservoir Flood Risk Index The Index is intended to provide a first cut screen for identification of sub-catchments in the LMB where further exploration and development of flood control use for a hydropower reservoir is of the highest priority.it also identifies which existing or proposed reservoirs have the greatest potential to be used for multiple use flood control.the Index can be used to answer three main questions: (iv) (v) (vi) 2.3 Design of the index Which reservoirs have the capacity to control floods, Which reservoirs are in high flood yield catchments, and Which reservoirs have the highest potential for downstream damage? Overview The Index looks at 67large hydropower dams across the LMB obtained from the MRC Hydropower Database.It aggregates three sub-indices for each of the large hydropower reservoirs: (iv) (v) (vi) Flood Control: what is the structural and technical capacity of the WSI to control flood events? This relates to indicators such as the reservoir emergency storage, regulating and discharge capacity. Natural Flood Threat 4 : what is the nature and magnitude of flood events in the subcatchments of the Mekong? This is generally measured in terms of flood frequencies and arrival times and is related to parameters such as the catchment slope, vegetation types and precipitation dynamics. Flood ImpactPotential: what are the land use and human habitation characteristics of the at-risk areas downstream of the WSI? This relates to indicators such as population, poverty, agricultural land area and ecological priorities. Based on these sub-indices, the Index utilises nineindicators to assess the Natural Flood Threat, Flood Control and Flood Impact potential characteristics for each of the 67 reservoirs (Figure 7). 3 Cumulative wet season rainfall can vary by +/- 30% between years (MRC, 2011) 4 When evaluating the Natural Flood Threat sub-index, the influence of dams in a cascade is not taken into account. It assumes the worst case scenario, where all dams are full and pass on the flow unimpeded (Section 5.1). 10

24 Tech note 1: Reservoir Flood Risk Index Figure 7: Indicators utilised in the Reservoir Flood Risk Index: (RED) Reservoir design and infrastructure characteristics which influence a reservoir s capacity to store flood waters, (GREEN) catchment characteristics that measure timing and magnitude of incoming flood events, (ORANGE) land cover and socio-economic characteristics of a 5km riparian corridor within 50km downstream of the dam structure. 1. Probable Maximum Flood (PMF) inflow yr ARI inflow 3. Catchment time of concentration 4. Reservoir emergency storage 5. Live storage vs mean annual flow 6. Discharge capacity vs flood inflow 7. Downstream riparian Population (and poverty levels) 8. Downstream riparian agricultural land 9. Downstream ecosystem services Figure 8 summarises the overall assessment process used to score each reservoir. Scores for each of the 9indicatorsare standardised into a rank 0-1 for each of the three sub-indices: (i) Natural Flood Threat, (ii) Flood Control and (iii) Flood Impact potential. The Natural Flood Threat and Flood Control scores are used to characterise the Downstream Flood Potential, which is then combined with the FloodImpactPotential to obtain the overall Reservoir Flood Risk Index score Ranking indicator scores The principles of Multi-Criteria Analysis (MCA) were used to integrate the indicators for Natural Flood Threat, Flood Control and FloodImpactPotentialsub-Indices into the Index. This method systematically ranks a number of diverse assessment indicators and provides a structured and transparent way of analysing complex issues (Engineers Australia, 2013). The process included the following: Standardisation was used for each indicatorof the Flood Threat, Flood Control and Impact Potentialsub-Indices. A score was determined from 0 to 1 so that they were considered on the same basis. The calculation was undertaken as follows (Annandale & Lantzke, 2000): RRRRRRRRRRRRRR = IIIIIIIIIIIIIIIIII VVVVVVVV ee MMMMMMMMMMMMMM VVVVVVVVVV iiii RRRRRRRRRR MMMMMMMMMMMMMM MMMMMMMMMMMMMM VVVVVVVVVV ( =rrrrrrrrrr oooo vvvvvvvvvvvv ) --- (Equation 1) The Natural Flood Threat and Flood ImpactPotentialsub-Indices had a consistent ranking where 0 is a low risk, and 1 is a high risk, whereas the Flood Control sub-index had 0 as a high risk and 1 as a low risk. Weighting was not applied to the rankings for the indicators, apart from one indicator of the Flood ImpactPotential sub-index, where the poor population is weighted higher than the non-poor. The other indicators equally contribute to the overall Natural Flood Threat, Flood Control and Flood Impact Potential ranking of the dam. Combining of the indicators for the Natural Flood Threat, Flood Control and Flood ImpactPotentialsub-Indices was undertaken by averaging the indicators together. The resultant value was then given a rank from Very Low to Very High by equally splitting the spread of results. 11

25 Figure 8: Development of the Reservoir Flood Risk Index: the Index uses standardised scores for 9indicators to estimate the Natural Flood Threat, Flood Control andflood ImpactPotential from reservoir releases. 12

26 SECTION B SUMMARY OF RESULTS & MAIN FINDINGS 13

27 3 Results of the Index Sections 3.1 and 3.2 below identify the analysis and main findings for the Natural FloodThreat and potential for Flood Control in Mekong reservoirs, which are then used to determine the Downstream Flood Potential (Section 3.3).Section 3.4 then presents the analysis and main findings for the FloodImpactPotential, which is used in combination with the Downstream Flood Potential to give the results of the Reservoir Flood Risk Index (Section 3.5). At the first stage of the development process, the Index can act as a screening tool for governments to set priorities on where and how flood control should be integrated into the project development cycle to maximise the benefit.the methodology used in determining Natural Flood Threat, Flood Control,Downstream Flood Potential, and FloodImpact potential is standardised to the sample set and so presents a scoring relative to the dams included in the assessment. Scores are therefore most useful in setting regional planning priorities.there were136 medium and large single-use hydropower projects under consideration across a number of national and regional planning frameworks. In addition, the Index can be used during the environmental review phase for designers to prioritise flood control in reservoirs where it is most needed. At the EIA review phase of project development, the Index will also allow affected communities and/or NGOs to gauge the capacity of individual reservoirs to control flood events based primarily on the relative reservoir size and capacity of the spillway gates as well as the potential benefits or damages to downstream communities and their assets. Last, the Index will allow hydropower developers to estimate the flood risk in a sub-catchment at the pre-feasibility and scoping phases of project development before firm agreements on electricity production and designs are finalised so that the importance/need for flood control is considered. Flooding and flood attenuation is highly variable amongst the sub-catchments of the Mekong and a function of a number of hydro-physical variables (rainfall recurrence intervals, catchment area, shape and time of concentration, stream geomorphology). Calculations of empirical formula for estimating peak flood events in ungauged catchments were developed and will be useful for hydropower operators attempting to establishthe size of design flood events in the data-scarce sub-catchments of the LMB. Some of the results presented below are dependent on the quality of the input data that was obtained from sources such as the MRC Database, Landscan and government statistics offices in different countriesthatcould not be validated.included in the results below are dams such as Nam Theun 2 and Xe Bang Nouan with very high flood control potential which has reduced the ranking of other dams. This is particularly relevant for the regulating capacity indicatorwhere the live storage was two to four times bigger than the mean annual flow. 3.1 Natural Flood Threat for hydropower contributing catchments Concept of Natural Flood Threat The threat of floods in a river is a function of (i) catchment precipitation dynamics and (ii) catchment physical characteristics that influence how rainfall passes through hydrological processes and discharges through the catchment s stream network. Catchment physical characteristics such as land use, shape, area, soil type, geology, forest cover and channel carrying capacity control how much water is infiltrated, how much is routed as runoff, how quickly water coalesces into the main channel and how fast the channelised flow travels (Figure 9). This studyidentified the most important characteristics influencing flood frequencies and magnitudes in areas of the Mekong Basin and used this to assess the flood threat for dams and downstream communities. The complexity and uncertainty of interactions that influence flooding means it is hard to predict the exact threat posed byfloods in any given year.to overcome the problem of uncertainty hydrologists and dam designers express flood threat in terms of probabilities including concepts such as: 14

28 Average Recurrence Intervals - a statistical benchmark used for comparison of floods. It is the expected number of years between exceedance of flood events of a given magnitude. For example if a dam catchment has a 50 year ARI flow of 1,000m 3 /s then it can be expected that a flow of 1,000m 3 /s will occur on average once every 50 years. It should be noted that this is a probabilistic estimate of what is likely to happen.in reality, it may be that a flood of this size occurs more often. Probable Maximum Flood (PMF) - the largest flood that could conceivably occur in a catchment. This is a very rare and unlikely event. The PMF does not have a designated return period but has been shown to be often in the range of 1.5 to 2.5 times the 1000 year ARI flood (GoN, 1990). Figure 9: Factors influencing catchment hydrological processes underlying the propagation and attenuation of flooding (SoQ, 2011) When designing a dam, designers must decide on a design flood peak inflow that guides the sizing of the reservoir and spillway. This decision is based on an analysis of the flood threat for the catchment and the level of acceptable risk of dam failure. If the design flood peak flow is exceeded it will likely lead to spillway damage, overtopping of the dam crest, potential dam failure and flooding downstream. The PMF represents the most conservative design peak inflow option andis adopted as the design peak inflow in situations where consequences of dam failure are unacceptable (FEMA, 2004). The design inflow for medium to large hydropower dams, where failure can have large and catastrophic impacts on the dam and downstream, is generally chosen as somewhere between the 100 year ARI to PMF. This study has calculated the 100 year ARI and Probable Maximum Flood for 67 dams in the Mekong Basin. This information was used to assess flood threat at two levels: (i) At the basin scale the information identifies which dams are in flood prone areas and are are therefore more likely to experience larger and more frequent floods; and (ii) At the individual dam level the adequacy of the design flood peak inflow can be assessed. A dam is more threatened by floods if its design peak flood inflow is lower than the calculated 100 year ARI or PMF. It should be noted that not all dams need to be designed to accommodate flows of 15

29 this magnitude. Designers of structures that are small or non-strategic may use a lower ARI flood if the impacts of dam failure at lower flows are acceptable. A third important factor for characterising flood threat is the response time that dam operators and downstream communities have to respond to a storm event that may cause flooding. A longer time to respond means that operators can prepare the dam by earlier releases and downstream communities can be warned to evacuate or prepare for a potential flood. A measure of this response time is the time of concentration which is the time needed for water to flow from the most hydraulically remote point in a watershed to the dam. The time of concentration varies throughout the basin and is a function of the catchment size, topography, geology, and land use. The results of analysis for the three indicators of the Natural Flood Threat sub-index are presented in this chapter: i) Indicator 1 - Catchment 100 year ARI flow and ratio of the 100 year ARI flow to the design peak inflow; ii) Indicator 2 - Catchment PMF and ratio of the PMF to the design peak inflow; and iii) Indicator 3 - Catchment time of concentration Overview of methodology for hydrological analysis in gauged catchments Calculating flood frequency The estimation of flood frequency is a complex problem that has led to the development of a number of different analysis approaches, each suitable to different situations and data availability. This study tooka flood frequency statistical analysis approach. This involved using discharge measurement station data, coupled with catchment size correction or regional regression, to transpose the measurement station flood frequencies to dam sites. A total of 123 discharge measurement stations can be found in the Mekong River Commission s hydrological database.after removing stations with insufficient or unreliable records, the number of useable stations drops to 52.Locations for these are shown infigure 10. Data for an additional three stations in the Sesan River basin in Vietnam were obtained from Power Engineering Consulting Company 1 (PEECC1). In most cases the 55 measurement stations do not match directly with the location of the 67 dams assessed in this study so methods were used to transpose the station results to the dam sites: One dam is gauged at the exact damlocation statistical flood frequency analysis was used to assess the flood frequencies at this site; 13 dams are located in gauged catchments in the same stretch of river as the discharge station at these sites statistical flood frequency analysis was used with a correction for catchment size; and 53 dams are located in ungauged catchments with stations in nearby catchments regional regression analysis was used for these dam sites based on statistical flood frequency analysis of the available stations. 16

30 Figure 10: Discharge stations, hydropower dams and hydrological analysis zones of the LMB: analysis of annual and seasonal rainfall was used to split main regions of hydropower development in the Mekong into 3 key zones: (i) Northern Lao-Thai zone, (ii) Central Lao PDR, and (iii) Central Highlands/3S region of Viet Nam, Cambodia and Lao PDR Gauged catchments statistical flood frequency analysis Natural Flood Threat in gauged catchments was calculated through statistical flood frequency analysis. The technique uses observed annual peak flow discharge data to calculate statistical information such as mean values, standard deviations, skewness, and recurrence intervals. This statistical data is then used to construct frequency distributions, which give the likelihood of various discharges as a function of recurrence. In this case the Log-Pearson Type III distribution was used.this distribution has been shown to be a good fit for events with return periods well beyond the observed flood events (OSU, 2005). 17

31 When a station was located in the same stream section as the dam location i.e. no major tributaries between the dam site and station the flood magnitudes were adjusted according to the relative catchment sizes Ungauged catchments regional regression analysis There is a significant gap of gauged data for tributaries in the Mekong Basin (Figure 10). Of the 67 dams assessed, only 14 had nearby stations that could be directly used for flood frequency analysis. For the 53 remaining dams, regional regression analysis was used to develop equations relating catchment characteristics to flood frequency. Figure 11: Overview of the regression analysis: approach used to transpose empirical relationships based on flood frequency and catchment characteristics to ungauged catchments for 53 of the 67 dams studies Catchment characteristics for hydrological stations obtained through GIS analysis Region 1 Catchment area Prediction equation for 100 yr ARI Catchment elevation Catchment slope Prediction equation for PMF Catchment compactness ratio Typhoon frequency Stream length Land use Divide stations into three regions Regression analysis to find the simplest and most accurate function of catchment characteristics that describes the 100 yr ARI and PMF Region 2 Prediction equation for 100 yr ARI Prediction equation for PMF Forest cover Flood frequency for hydrological stations obtained using Log Pearson Type III distribution Region 3 Prediction equation for 100 yr ARI 100 year ARI Prediction equation for PMF PMF A detailed description of the method has been developed as a separate briefing note (MK12 Technical Note 2: ICEM, 2014) and a summary is provided here. A key first step in the regional regression analysis was to identify areas of the basin where there was insufficient coverage of stations to undertake regression analysis and then separate the remaining area into regions of similar orographic and precipitation characteristics (Figure 10). Areas of similar orographic and precipitation characteristics will exhibit similar flood frequency relationships. Physical characteristics of the catchments above the station sites in each region were then statisticallyrelated to the 100 year ARI flow. This analysis resulted in an equation for each region that can be used to estimate the 100 year ARI flow for any location within the region. The resulting regional regression equations were used to calculate the 100 year ARI and PMF flows in cases where adequate data was not available at or near the dam site Indicator year ARI flow and ratio to peak design inflow The 100 year ARI peak flow is the peak flow expected to occur on average every 100 years. It can also be expressed as having a 1% chance of being exceeded in any given year. The 100 year ARI flow is important in dam design as it is often used as a benchmark standard for design peak inflow. 18

32 Regression analysis found that key catchment characteristics influencing the 100 year ARI in the Mekong Basin were the catchment area, catchment total stream length, typhoon frequency and catchment compactness ratio. Table 1 presents the empirical equations derived to estimate the size of the 100 year ARI in ungauged catchments together with the correlation coefficient obtained from gauged catchment data. Table 1: Empirical equations for predicting 100 year ARI flows in the LMB by climate zone Region Equation R 2 No. stations 1 Northern Annamites and North Eastern Thailand ARI 100 = St A 0.23 SL Central Annamites ARI 100 = A A St The Three S basins ARI 100 = A SLK c Where: A = Catchment area; St = Number of typhoons in period ; SL = Total stream length in the catchment; K c = catchment compactness ratio (K c = X Perimeter/Area 0.5 ) The 100 year ARI flow was calculated for 67 of the dams contained in the MRC database. In addition, the ratio of design peak inflow - obtained from the MRC database - to the 100 year ARI was calculated for each dam. If this ratio is greater than one then the 100 year ARI flow is larger than the design peak inflow, indicating that the dam is not designed to accommodate the 100 year ARI flow. Main findings The results show that 40% of the 67 dams have a design peak inflow of less than the 100 year ARI flow. These dams are at higher risk of failure if high flow events occur. Xelabam Dam in Lao PDR has by far the worst ratio of 100 year ARI to design peak inflow, indicating that its design peak inflow was based on a much lower ARI flow. This may be because it is a small scale dam which may not cause significant impacts if it fails. The remaining 60% of dams have design peak inflows of more than the 100 year ARI flow and are therefore at lower risk of failure due to the 100 year event. The Nam Hinboun 2, Nam San 3B, Nam Chian and Xe Lanong 2 dams have design peak inflows of well above the 100 year ARI flow, meaning they are at a lower risk of failure due to high flows. The largest 100 year ARI flows are shown to occur at the Lower Sesan 2 dam. This dam, with by far the greatest catchment area of the dams studied, has a 100 year ARI flow of more than double the next ranked dam. The design peak inflow for the Lower Sesan 2 is also very high, more than double the 100 year ARI flow, whilst the spillway discharge capacity is one and a half times larger. Given however that the dam is at the bottom of two large river basins that experience extreme flood events, the spillway should have a larger discharge capacity than it currently does. 19

33 Figure 12: Analysis results for 100 year ARI: the red dotted line dams with a design 100 year ARI ratio above the line have a design peak flow of less than the calculated 100 year ARI Indicator 2 - Probable maximum flood For dam designers and operators, the PMF is an important consideration as it is the largest possible flood threat that may feasibly occur at a dam site given the geographic, meteorologic and hydrologic characteristics of the catchment. Calculation of the PMF does not depend on statistical analysis of flow records, but on determining the most adverse, plausible meteorological conditions that could be expected to occur in the basin. A large amount of detailed basin specific information must be collected in order to calculate the PMF for any given catchment. For basin-level and pre-feasibility assessments these data requirements are excessive. Therefore this study applied an estimate that is often used for PMF evaluations when data or time is limited. A study on PMF and the 1000 year ARI around the world showed a range in ratios of 1.34 to 2.94, therefore the study has chosen a factor of two for estimation of the PMF based on the 1000 year ARI (GoN, 1990). PMF in gauged catchments was estimated through statistical flood frequency analysis to calculate the 1000 year ARI flow and then multiplying it by a factor of 2. When a station was located in the same stream section as the dam location i.e. no major tributaries in between the dam site and station the PMF was adjusted according to the relative catchment sizes. For ungauged catchments regional regression was undertaken to link catchment characteristics to the PMF. This analysis resulted in an equation for each region that can be used to estimate the PMF for any location within the region (Table 2).The analysis found that the catchment characteristics that were key determinants of the PMF included typhoon frequency, catchment total stream length, catchment average slope and catchment area. The resulting regional regression equations were used to calculate the PMF in cases where adequate flow information was not available at or near the dam site. Note that this PMF estimation is developed for basin-wide and feasibility level assessments and should not be used for specific dam design purposes. If detailed design were being undertaken then a more thorough study of available rainfall data would need to be conducted for the dam site. 20

34 Table 2:Empirical equations for predicting Probable Maximum Flood (PMF) in the LMB by climate zone Region Equation R 2 No. stations 1 Northern Annamites and North Eastern Thailand PPPPPP = SSSS AA Central Annamites PMF = A 1994 Sl 1 Northern Annamites and North Eastern Thailand St PMF = A 2.4SL Where: A = Catchment area; St = Number of typhoons in period ; SL = Total stream length in the catchment; Sl = Catchment average slope Main findings The PMF was calculated for 67 of the dams located in the MRC database. In addition the design peak flow for each dam, obtained from the database, was compared to the PMF to provide an indication of which dams have been designed to accommodate the PMF. The results are presented in Figure 13 below. The catchments of the Lower Se San 2 and Lower Sre Pok 2 dams have the largest PMF due to their large catchment size. Seven of the top 10 PMFs are located in the Three S basin including on the Sesan, Srepok and Sekong Rivers. This result is to be expected because these catchments exhibit some of the highest flows of the basin and provide more than 20% of the total Mekong River mean annual flow (MRC, 2009). Themethodology used for assessing Natural Flood Threat does not account for cascade operations. For example, the Se San 4A is ranked with a high PMF but the dam is a run-of-river re-regulation dam situated directly below the Se San 4 dam. Its PMF is therefore directly related to the management of the Se San 4 dam and it has not been designed to provide flood control services. The study team was not able to identify which dams act as re-regulation dams and the dams for which flood threat is directly related to the dam or cascade located upstream. 21

35 Figure 13: Analysis results for PMF: Note the red dotted line dams with a design flow/pmf ratio above the line have a design peak flow of less than the calculated PMF The results show that 65% of the dams have a design peak inflow lower than the PMF indicating that they are at higher risk of failure during extreme floods. The 35% of dams designed for flows of higher than the PMF are of lower risk of failure due to unexpectedly high flows. It is not surprising that a large percentage of the dams are not designed for the PMF as this is the largest flood possible in the catchment and dam designers generally use a lower design flow and accept the slight risk that this poses. Xelebam Dam was again most at risk which is likely due to the small risk of impact of downstream releases associated with flood response. The best performing dams for this indicator were Nam Mang 1 and Nam San 3B. These dams had design peak inflows of well above the PMF flow, meaning they are very unlikely to experience a flood event for which they are not designed Indicator 3 - Time of concentration The time of concentration, the time it takes for runoff to travel from the most hydraulically remote point to the dam site, gives an indication of how fast a flood may occur after a heavy rainfall event falls on the dam catchment. This is an important measure of the response time that dam operators and downstream communities will have to prepare for a flood. A longer time to respond means that operators can prepare the dam by earlier releases and downstream communities can be warned to evacuate or prepare for a potential flood. The time of concentration provides a comparable estimate of the speed at which floods occur in a catchment. When the duration of a storm exceeds the time of concentration, the indicator provides a measure of the timeto peak discharge i.e. the period of time between the start of the storm and the flood peak. In situations where the duration of the storm event is less than the time of concentration, the indicator doesn t give the exact time to the flood peak but still provides a good comparative indication of the flood response time. 22

36 Figure 14: Different flow types contributing to the time of concentration (modified from Ghelardi, 2011) Time of concentration varies with the size and shape of the catchment, the land and channel slope, the land use, the intensity of rainfall and whether flow is overland or channelized (Ghelardi, 2011). The flow paths for runoff can be split into three categories: overland, shallow concentrated and concentrated. In the upper reaches of a catchment overland flow occurs as shallow sheet flow, typically at a depth of 20-30mm. Within a short distance, typically around 100m, the water collects into shallow concentrated flow in rills and swales at depths of mm. As flow continues to accumulate it concentrates into larger and deeper channelsuntil it eventually collects into a river channel.figure 14illustrates a typical subdivision of these three different flow paths.the time of concentration is the sum of the travel times in all three flow paths (Ghelardi, 2011). There are numerous methods for calculating the time of concentration for a catchment. More complex methods use detailed catchment characteristics such rainfall intensity and detailed land use information. This study is developed to be easily replicable at the basin level and pre-feasibility stage and therefore utilised a simple equation developed specifically for the Mekong Basin and for which the parameters could be calculated using GIS techniques (Tussaporn and Mongkolsawat, unknown):(equation 2).The format of the formula is similar to the commonly used Kirpich and Bransby-Williams equations for calculations of time of concentration except with a change of the coefficients to match the regional conditions of the Mekong. TT cc = LL HH (Equation 2) Where T c = time of concentration L = Length of travel (ft) which is the total length of overland flow, shallow concentrated flow and channelized flow H = Slope of travel which is the slope from the most hydraulically remote point to the dam site travelling along the runoff route 23

37 In this study it was not possible to independently validate the coefficients used in the above equation. Therefore the results for time of concentration should be used as a relative indication but not as a measure of absolute time of concentration values. Main findings The time of concentration was calculated for 67 of the dams held within the MRC database. The indicator was found to vary widely throughout the basin by a factor of 175. The results are presented in Figure 15. Dams with large catchment areas scored well for this indicator. The Lower Se San 2 damhas the largest catchments of the dams analysed and were shown to have relatively much higher time of concentration - largely due to a long flow travel distance from the hydraulically most remote point to the dam site of more than 400km. A quarter of the dams analysed showed a low time of concentration. These dams are more highly threatened by floods due to the short time available for dam operators and downstream communities to prepare for the flood impacts. The catchments of the Nam Hinboun 2, Nam Phak, Nam San, Nam Mang 3 and Nam Pay performed the worst for this indicator. These dams are located in the northern and central Annamites and have small steep catchment areas which tend to result in shorter hydrological response times. Figure 15: Analysis of results for time of concentration: (bar graph) standardised rank for flood risk association with time of concentration; (scatter plot) time of concentration for LMB dam catchment in days Conclusions The natural flood threat component of the Index utilises information on the hydrological characteristics of the reservoir s contributing catchment to rank the flood risk of inflows to the reservoir. The component utilised three standardised ranked flood threat indicators: (i) 100 year 24

38 Average Recurrence Interval (ARI), (ii) Probable Maximum Flood (PMF), and (iii) time of concentration 5. These three parameters were combined to calculate the Natural Flood Threat. The sub-index results indicate the dams most threatened by large natural flood events and does not account for dam design or cascade effects. The results of the top and bottom ten dams ranked for Natural Flood Threat are shown in Table 3and Table 4below, with the main findings including: 1. The 3 S catchments have some of the highest flood risks in the LMB:Five of the top ten most threatened dams are located in the Three S catchment area. This area tends to have larger catchments and is an area of high precipitation, which leads to higher flows buta longer time of concentration. 2. Northeast Thailand and the Northern Annamite catchments have some of the lowest flood risks in the LMB:Nine of theten dams least threatened by floods are located in the northern Annamites and north-east Thailand, where the catchment areas and rainfall volumes are comparatively smaller, leading to lower flows. Table 3: Highest relative Natural Flood Threat: Reservoir catchments with the highest NaturalFlood Threat RANK NAME 1 Lower Se San2 + Lower Sre Pok 2 2 Xe Kong 3d 3 Xe La Bam 4 Xe Kong 3up 5 Xe Kong 4 6 Xe Don 2 7 Xe Kong 5 8 Xe Kaman 1 9 Xe Xou 10 Xe Kaman 2A Table 4: Lowest relative Natural Flood Threat: Reservoir catchments with the lowest NaturalFlood Threat RANK NAME 58 Nam Pouy 59 Nam Khan 3 60 Nam Poun 61 Nam Ngum 4A 62 Nam Ngum 2 63 Buon Kuop 64 Nam Suang 1 65 Nam Khan 2 66 Nam Ngum 1 67 Nam Khan % of the dams assessed have design peak inflows lower than the 100year ARI flow. Thesedams are at higher risk of failure if high flow events occur. 4. A quarter of the dams assessed were built in catchments with quick response times, which means operator response time is also short. 5 Time of concentration refers to the time required for surface water travel from the most upstream point in the catchment to its outlet at the reservoir headwaters. 25

39 3.2 Flood Control Potential for hydropower reservoirs A dam can provide flood control, or mitigation, by temporarily storing flood water and discharging it slowly. This reduces flood peaks and flood frequency on the river, as well as the incidence of coinciding peaks with other rivers in the catchment downstream of the dam (ICOLD, 2013; Bradley, 2012). The measure of the ability of a dam to do so is an important component of a Reservoir Flood Risk Index. The flood control characteristics of a reservoir relate primarily to its management, physical capacity to store and discharge the water, and its location in the catchment. Typically, reservoir management in the Basin does not account for allowance of a flood storage volume during the wet season whether it was designed for it or not, as the water level is kept as high as possible for maximum electricity generation potential. Flood control then becomes solely related to the reservoir s physical characteristics and its location in the catchment. A reservoir s physical capacity for flood control is generally dependant on the: Discharge capacity of its outlets:spillway type and discharge capacity, bottom outlet discharge capacity, turbine discharge capacity Storage volume:available storage volume above the dam FSL, and the live storage volume. Dams should be carefully designed to store and discharge water of a volume and flow depending on catchment characteristics (Section 3.1), as improper sizing can lead to potential dam failure that can cause extensive loss of life, infrastructure damage and ecological destruction. The Mekong River Commission s (MRC) Hydropower Database for the LMB contains detailed information for 136 large hydropower dams in the LMB, including discharge and storage data. Information on each characteristic however was not available for every dam. The analysis was therefore limited to dams for which all data was available, which came to 67 of the original 136 dams (Figure 5). Results of analysis for the three indicators of Flood Control sub-index are presented in this chapter: (i) Indicator 4 - Reservoir emergency capacity to store flood water, (ii) Indicator 5 - Live storage versus Mean Annual Flow (regulating capacity), and (iii) Indicator 6 - Dam discharge capacity Indicator 4 - Reservoir emergency storage capacity This indicatorcompares the available emergency storage volume for each dam (defined to be within a 1.0m freeboard allowance between FSL and the Dam Crest Level) to the 100 year ARI inflow volume over a 24-hour period. The most important flood control function a reservoir has is to store water during a flood and reduce the magnitude of downstream peaks. Given that reservoir management in the Basin involves keeping water levels high in the wet season, the available storage is usually very limited. Reservoirs typically have flood storage available between the FSL and the Maximum Water Level, which can also be referred to as the spillway surcharge storage (Figure 16). This surcharge storage is used when a spillway constricts the flow, causing water to back up, or surcharge, before passing through the spillway. This, along with the freeboard buffer, could be considered as emergency storage in the event of large and unmanageable flood inflows that would require discharging as soon as possible to prevent dam safety issues. 26

40 Figure 16: Typical reservoir storage levels: (Image source: Maximum Water Level (MWL): is the maximum level to which the water surface will rise when the design flood passes through the spillway (Mohanty, 2012). Full Supply Level (FSL): normal maximum operating water level of reservoir when not affected by floods. This water level corresponds to 100% capacity. Dam Crest Level (DCL): is the top of the dam and the point at which overtopping could occur which would risk dam failure. Low Supply Level (LSL): normal minimum operating level of reservoir when seasonal drawdown is at maximum. Freeboard: is the minimum vertical distance between the reservoir MWL and the DCL, and is designed to protect dam integrity from overtopping and dam failure caused by large wind induced waves, landslide or earthquake effects or settlement (Ontario Government, 2011). The 100 year ARI flood inflow is commonly used in dam design and can result in very large volumes of water entering a reservoir over a short period of time. Nam Theun 2 Dam (Lao PDR), for example, experienced this level of inflow beginning during the night, leading to a rise in the water level of 3.5m over 24-hours. Dam operators, who were resting at that point, were left with very limited time to respond. Therefore a dam s ability to provide emergency storage of the 100 year ARI inflow over a 24-hour period is a good measure of its flood control capacity. Main findings The analysis used the freeboard and reservoir surface areas for each dam. The information provided in the MRC hydropower database assumes a 1.0m freeboard between the FSL and the DCL, which was used to determine the maximum storage area and mean storage depth for each dam for comparative purposes 6. As the freeboard is assumed the same, the dams with the greatest storage capacity have the largest reservoir surface areas. The 100 year ARI inflow used for each dam was calculated as per the methodology followed in Section The results are shown in Figure 17, ranked from lowest to highest available emergency storage capacity, where the higher numbers are for dams with greater capacity to store the 100 year ARI inflow over 24-hours. The results show that the top ten dams account for approximately 54% of the total emergency storage capacity for all 67 dams. Correlation between emergency storage capacity and dam height is not as strong as expected. The majority of below average height dams also have below average emergency storage capacity, however, many of the highest dams have lower emergency storage capacity compared to medium sized reservoirs. The lowest capacity dams are quite low in height, being O Chum 2 (10.0m height), Xeset 1 (18.0m height) and Dray Hlinh 2 (7.0m height) Dams.The highest are low to medium sized, being Xe Neua (28.0m height), Xe Bang Nouan (65.0m height) and Nam Ngum 1 (75.0m height) Dams. The dams at particular risk are those with low capacity to store water and a relatively large volume of live storage (water stored between FSL and LSL). In the event of dam failure, resultant inundation 6 MRC hydropower database does not indicate whether the MWL has been included in this freeboard or not.in reality, it is likely that: (i) the MWL is below this freeboard, and (ii) the freeboard is bigger than 1.0m. This is however a conservative estimate. 27

41 from this live storage volume would cause damage and potential loss of life to communities. Dak E Mule and Nam Ngum 5 Dams are both examples of high dams (>100 m) with large live storages ( mcm) but comparably very small capacity to store emergency floodwater. These disproportionately large live storages could, under human error or mismanagement, result in downstream flood flows greater than natural conditions would create. Figure 17: Reservoir emergency storage - Standardised ranking: (0 0.2) 44 dams are in the very low category; ( ) 17 dams are in the low category; ( )2 dams are in the medium category; ( )2 damsare in the high category; ( )2 dams are in the very high category Indicator 5 - Regulating capacity (live storage vs mean annual flow) This indicator looks at the ratio between each dam s live storage and the Mean Annual Flow (MAF) from the river, which is a conventional measure of the reservoir s potential regulating capacity. A hydropower reservoir s storage capacity (live storage) is its most important characteristic, as this removes the fluctuations in supply and allows for base load electricity generation. The live storage of a reservoir depends mostly on the topography of the site and the height of the dam. In order to fully understand the available volume, engineers conduct topographic surveys and use GIS techniques to prepare contour maps that allow the development of a stage-storage and stage-area relationship (Figure 18). This enables a designer to determine the potential volume of the reservoir and the water spread area at different elevations up to the dam height (IIT Delhi, 2013). In addition to storage, hydrological investigations are also undertaken to study runoff patterns and estimate yield of the river from season to season and year to year, normally defined as the mean annual flow (MAF) at the reservoir inlet. Reservoirs can have sufficient operating, or live, storage to accommodate for a few seasons of low flows, and for normal catchment yield throughout the year. The relationship of live storage to mean annual flow relates the storage capacity to the inflow capacity of the reservoir and is closely linked to the reservoir s residence time. It is a measure of the reservoir s regulatory capacity under normal operating conditions and can provide an indication of the operational flexibility with which flood water could potentially be managed. 28

42 Figure 18: Stage-storage and stage-area relationships (Source: IIT Delhi, 2013) Main findings The analysis used live storage and MAF figures from the MRC database. The live storage figure was divided by the MAF to give the regulating capacity of the dam. The dams with a standardised ranking closer to 1.0 have a higher regulating capacity. The results are shown in Figure 19 below. Figure 19: Live storage vs MAF - Standardised ranking: (0 0.2) 61 dams are in the very low category; ( ) 4 dams are in the low category; ( ) 1 dam is in the medium category; ( ) no dams are in the high category; ( ) 1 dam is in the very high category. Results show that there are three dams with live storages that are much bigger than MAF compared and can be said to have a high capacity of regulation: (i) Xe Bang Nouan, (ii) Nam Theun 4 and (iii) Nam San 2. Xe Bang Nouan s live storage is over four times the MAF. 29

43 Dams with very low regulating capacity include Dray Hinh 1 and 2, Sesan 3 and Xeset 1, which are located in southern Lao PDR and the 3S basin. Figure 20 shows a comparison of the potential regulating capacity of selected high profile dams and cascades (a series of dams on the same river) overlaid with average annual precipitation in the LMB. The table in Figure 20 also shows the frequency of tropical storms that have crossed the catchment of the dam using over 50 years of storm track data (OCHA, 2013), which is a proxy measure of the occurrence of extreme rainfall and subsequent high inflows into the reservoir. The results show that Nam Suang 2 (in Lao PDR) has a high regulating capacity and is located in an area of low precipitation (1,400 to 1,600 mm) in a catchment that has been directly impacted by tropical storms only once every 11.2 years. It is therefore quite low risk. Xe Kong 3up (Cambodia) however has a low regulating capacity, is located in a high precipitation catchment (2,000 to 2,200mm), and is impacted by tropical storms once every 1.9 years, which is very frequent and therefore is high risk. The Lower Sesan 2 dam (Cambodian central highlands) was found to have a low regulating capacity, which is in contrast to its high emergency storage potential. This dam is located in a moderate precipitation area (1,600 to 1,800mm), is impacted by tropical storms every 2.7 years, and is considered as high risk. The Xe Kaman cascade dams all have a regulating capacity of 10% or less and are located in a high precipitation catchment which is regularly impacted by tropical storms. These dams are at high risk. This indicator shows that of the dams measured, over half have a regulating capacity of less than 10%. In theory this can be compensated for through the provision of a buffer storage, however, given that management practices in the LMB rarely account for flood buffer storage, the risk becomes more significant. Figure 20: Examples of dams with low, medium and high regulatory capacity Indicator 6 - Discharge capacity 30

44 This indicator looks at the ratio between the spillway discharge capacity and the 100 year ARI flood inflow to determine the dam s capacity to discharge extreme inflows. The discharge capacity of a dam is the maximum amount of water it can safely release to reduce the water level when there is a flood risk to the dam. Hydropower dams typically have three ways to release water, which are as follows: 2. Spillway (Box 2): This includes both the service and auxiliary/emergency spillways. A dam s spillway is sized for a design flood peak inflow that is determined through a detailed hydrological analysis. Whilst this design peak varies depending on the dam s hazard categorisation, a good minimum measure to use is the 100 year ARI inflow. The spillway should have some constricting characteristics as this improves flood routing through the dam and can create a required design head to drive the discharge, however, the spillway should be able to pass flood inflows quickly to avoid dam safety issues. 3. Low level outlet: The low level outlet is usually a piped outlet with an invert level that is much lower than the spillway, and in most cases below the LSL to provide sufficient head required to achieve the design discharge (IIT Kharagpur, 2013). This outlet (Figure 21) provides functions such as (i) release of minimum environmental flows, (ii) emergency discharges for reservoir drawdown (i.e. during flood), (iii) controlled releases required during impoundment of a reservoir to control the rate of filling, which allows for monitoring of the dam and foundations under the increasing load and pore pressure, and (iv) to bypass or divert flows which may be required during dam construction or spillway maintenance works. This outlet can be used during floods provided that high tailwater levels from flood flows or other spillways do not impede its operation. (Ontario Government, 2011). 4. Turbine discharge: The turbine discharge is the maximum amount of water that can be passed by the powerhouse, and depends on the size and number of turbines. During a flood event however, the powerhouse would likely lower output or shut down turbines for four main reasons (Ontario Government, 2011): (i) The plant will be operated depending on requirements of the power system and the demand may be less during a storm, (ii) There may be damage to the electrical transmission system that will require repair, (iii) High levels of debris including sediments and tree branches in the water could damage plant, and (iv) Tailwater levels from the flood could inundate the powerhouse turbine floor. Box 2: The spillway and its role in dam safety Source: US Army Corps of Engineers Digital Visual Library Source: Missouri University of Science and Technology The spillway plays a critical role in dam safety by ensuring that any excess water can be discharged from the storage in a timely fashion to prevent overtopping and dam failure. Dams without specific flood mitigation measures can still attenuate floods purely through the constricting effect of the spillway, also known as flood routing, as water tends to back up above the spillway crest and the flood discharge peak becomes reduced (Bradley, 2012). The sizing depends on the catchment properties and the hazard category (Knoop, 2011). There are many different types of spillways, including service (always in use) and auxiliary (rarely used unless in case of large floods that exceed the service spillway discharge capacity) and they can be built as part of the dam or a separate structure such as for embankment dams (Mazumder, 2011). The main components of a spillway include (Rajasthan Government, 2000): Entrance channel: admits and controls rate of water coming into the spillway to be discharged, Conduit: carries discharge from the entrance structure to the low level outlet downstream, and Outlet structure: dissipates the high-energy discharge flows to avoid excessive erosion of the channel bed and conveys the water to the downstream channel. The major differentiating feature of a spillway is whether it is controlled or uncontrolled. Controlled spillways have a gated structure that can be closed to prevent discharge, or opened incrementally to release water depending on the flow rate required. For uncontrolled spillways, the elevation determines the reservoir s FSL, and free flow discharge is initiated when reservoir levels surpass the spillway s crest level. Gated spillways will generally allow earlier response, more control and higher permissible reservoir levels than overflow spillways allowing for a great control of flood 31 releases if managed for that purpose. Therefore the spillway characteristics of a dam determine its capacity to control floods and enhance dam safety.

45 Figure 21: Low Flow Outlet (Source: MWH Global) Therefore this indicator assumes that under flood conditions, a dam s discharge capacity will only comprise the service and auxiliary/emergency spillway capacities, because the turbine and low flow outlets cannot be safely operated.the dam spillway discharge capacity was divided by the calculated 100 year ARI inflow (Section 3.1.3) and standardised to obtain a ratio. Main findings The results show that 88% of the 67 dams have a spillway discharge capacity that is greater than the 100 year ARI flood inflow, and 1% has a spillway discharge that is equal. These dams have a relatively lower risk. The remaining 11% have discharge capacities that are lower than the 100 year ARI flood inflow and therefore have higher risk.this is a relatively small proportion of the dams assessed, however, it is important to note that the 100 year ARI inflow may not be the design flood peak inflow for all dams. Setting the design inflow depends heavily on catchment hydrology and the hazard category of the dam, which is determined by the potential for loss of life as well as environmental and economic damage downstream. The projects with the best potential to control flood flows with their spillways include: Nam Poun and Buon Kuop dams, which have very large spillway capacities relative to the flood inflow. Dams such as Se San 4A and Se San 3 also have very good ratios which put them at low risk, which is the opposite of the findings for regulating capacity which showed them having little potential. The dam with a discharge capacity more or less equal to the flood peak inflow was Xe Xou Dam, in Lao PDR. 32

46 Figure 22: Flood Inflow versus dam discharge - Standardised ranking: (0 0.2) 35 dams are in the very low category; ( ) 18 dams are in the low category; ( ) 10 dams are in the medium category; ( ) 2 dams are in the high category; ( ) 2 dams are in the very high category Conclusions The flood control component of the Index assesses what is known about the design of the 67 reservoirs to rank the reservoirs by their potential to control flood peaks, assuming they would be managed for that purpose. General design information was used for spillway sizing, reservoir dimensions and design turbine flows to assess the capacity of the dams to store and discharge flood waters. The three indicators used were: (i) Reservoir emergency capacity to store flood water, (ii) live storage as a proportion of mean annual inflows, and (iii) dam discharge capacity. The results of top and bottom ten dams ranked for Flood Control are shown in Table 5andTable 6 below. Table 5: Highest relative Flood Control: Reservoirs with the greatest potential to manage flood events RANK NAME 1 Xe Bang Nouang 2 Nam Hinboun 2 3 Nam Poun 4 Nam Ngum 1 5 Nam Mang 1 6 Nam Mang 3 7 Nam San 2 8 Nam Theun 4 9 Buon Kuop 10 Nam Theun 2 33

47 Table 6: Lowest relative Flood Control: Reservoirs with the least potential to manage flood events RANK NAME 58 Xe Kaman 2A 59 Xelabam 60 Dray Hlinh1 61 Dray Hlinh 2 62 Xe Katam 63 Nam Kong 2 64 Nam Pay 65 Nam Phak 66 Xeset 1 67 O Chum 2 Main findings included: 1. In the LMB some of the largest dams have some of the smallest relative capacity to store extreme flood flows in emergency situations. The dams at particular risk are those with low capacity to store water and a relatively large volume of live storage (water stored between FSL and LSL). In the event of dam failure, resultant inundation from this live storage volume would cause damage and potential loss of life to communities. Dak E Mule and Nam Ngum 5 Dams are both examples of high dams (>100 m) with large live storages ( mcm) but comparably very small capacity to store emergency flood water. These disproportionately large live storages could, under human error or mismanagement, result in downstream flood flows greater than natural conditions would create. 2. More than half of the dams assessed have a regulating capacity of less than 10%. In theory this can be compensated for through the provision of buffer storage, however, given that management practices in the LMB rarely account for flood buffer storage, the risk becomes more significant % of the 67 dams have a spillway discharge capacity that is greater than the 100 year ARI flood inflow, and 1% has a spillway discharge that is equal. These dams have a relatively lower risk. 3.3 Downstream Flood Potential: Combined Natural Flood Threat and Flood Control ranking The Flood Controland Natural Flood Threatsub-Indices were combined to create a Downstream Flood Potential Index. A summary matrix of all dams considered and their relative scores is presented in Figure 23 below. 34

48 Figure 23: Downstream Flood Potential scores for Mekong hydro-electric reservoirs The dams ranked as high and low for combined Natural Flood Threat and Flood Control potential are shown in Table 7 and Table 8below. The following conclusions were drawn in terms of downstream flood potential: 6. The downstream flood potential for LMB reservoirs is exacerbated by poor capacity in existing hydropower designs to be utilised for flood control. For the majority of reservoirs assessed, most have low or very low capacity to regulate flood inflows. This means the majority of reservoirs are limited in their capacity to provide flood control for downstream areas because the incorporation of flood control as part of reservoir operations will require a redesign of reservoir and spillway infrastructure. 7. Dams located in medium, high and very high Natural Flood Threat areas of the Mekong Basin are generally not well designed for flood control. This is an important finding as it indicates that the first change needed to improve reservoir flood control in the region is improving the structural design of reservoirs. 8. The planned and existing reservoirs in the 3S basins have the highest down stream flood potential, largely due to the combination of poor design capacity for reservoirs to accommodate flood inflows and the large magnitude of flood events in the 3S catchments (Figure 24). 9. Lower Sesan 2 Dam and Xe Kong 3d exhibit the highest downstream flood potential of all LMB hydropower projects.both are smaller dams located at the bottom of a cascade, and therefore are likely to be regulating storages which is why they are have low flood control capacity. 10. Three reservoirs in the Xe Kong cascade (3D, 3up and 5) were identified as very high and high risk, due to a lower flood control capacity and highercatchment flows.five reservoirs in the Xe Kaman cascade (2A, 2B, 3,4A and 4B) were found to be of high flood risk primarily due to their poor flood control (small live storage and emergency storage). 11. Of the five reservoirs ranked as low risk, all have medium to high flood control and are located in small catchments with low flows. 35

49 Table 7: Downstream Flood Potential priorities: Very high and high ranked FLOOD RISK NAME Very High Lower Sesan 2 Very High Xe Kong 3D High Xelabam High Xe Kong 3up High Xe Kong 5 High Xe Kaman 2A High Xe Kaman 2B High Houay Lamphan High Dak E Mule High Xe Kaman 4A High O Chum 2 High Xe Kaman 4B High Nam Phak High Nam Kong 2 High Xe Kaman 3 High Xe Katam Table 8: Low Downstream Flood Potential FLOOD RISK NAME Low Buon Kuop Low Nam Ngum 1 Low Nam Poun Low Nam Hinboun 2 Low Xe Bang Nouan 36

50 Figure 24: Geographical distribution of downstream flood potential for LMB hydropower projects: Highest flood potential is in the 3 S basins 37

51 3.4 Floodimpact potential for hydropower reservoirs The final sub-index of the Reservoir Flood RiskIndex is an assessment of the potential of impacts to human and ecological assets in the riparian zone immediately downstream of the dam gates. The parameters used in the analysis are: (i) Population, (ii) Poverty, (iii) Land area with an agricultural use component (10 parameters) (iv) Key Biodiversity Areas (KBAs) and (v) Protected Areas. These parameters were combined to create the three indicators of the Flood Impact Potential Sub-Index: Population, Agriculture and Ecological Priority. The information was obtained from a number of sources, shown in Table 9. Table 9: The Flood Impact Potential indicators and their source data INDICATOR Population DATA SOURCE OF GIS DATA AND MAPPING LAYERS LandScan 2011 Global Population Poverty General Statistics Office Cambodia 2008, Lao PDR 2006, Viet Nam 2010, Thailand 2010 Agricultural area International Food Policy Research Institute 2000 GRID 1 km pixel size Official/Global Key Biodiversity Areas Protected Areas Mekong River Commission Mekong River Commission The Population Indicator combines poverty level data with the population data 7 to give a weighted population score; the Agriculture Indicator consists of ten land use parameters; and the Ecological Priority Indicator consists of Key Biodiversity Areas (KBAs) with Protected Areas. In general, the Population Indicator and the Agriculture Indicator are correlated, while both are negatively correlated with the Environment Indicator. The following sections describes the parameters and Indicators, and briefly presents the results with a map of the geographic distribution of the Indicator, a list of the the top five scoring dams, and a graph showing the scores. The full Impact Potential Sub-Index ranking (obtained from averaging the 3 Indicator scores) and the map of the Potential Impact scores is presented in the Appendix Use of the Zone of Influence The analysis required defining a boundary, or limit, of the influence area for each hydropower reservoir in order to collect information on the parameters. This was done using the Zone of Influence piloted by ICEM for the Asian Development Bank s Strategic Environmental Assessment (SEA) of the GMS Power Development Plan (ICEM, 2013b). The use of the Zone of Influence was acknowledged to represent a significant simplification of the complex hydrological, environmental and socio-economic reality of dam impacts, but was judged by experts to usefully demonstrate differences between power plan scenarios at the aggregate level, rather than to express net impacts for any particular dam or set of dams. For the SEA, the methodology was accepted by the Asian Development Bank and representatives of the Regional Power Trade Coordinating Committee (RPTCC). Because this index represents a ranking of the relative impacts of dams, rather than the net impacts of each dam, the Zone of Influence remains a useful if imperfect tool. The downstream Zone of Influence is a corridor that runs along the river alignment, beginning from the dam location and stretching 50 km downstream (see Figure 28 and Figure 31). The width is 2.5 km either side of the channel for a total of 5 km. Since detailed flood modelling is not available, this was chosen based on the expert judgment of environmental and social experts of the SEA on the assumption that communities within 2.5 km on each side of the river useit on a regular basis and this 7 Producing two parameters country defined poor and non-poor totals. 38

52 usage is at direct risk of floods. This represents a danger not only to their livelihoods and property, but also to their physical well-being. The 50 km length was chosen to represent the distance downstream within which variations of peaking discharges from the dam is moderated and where there would only be a slight variation between daily peak and low flows. In reality this variation would depend on the characteristics of the river valley and other tributaries.furthermore, additional variation between daily peak and low flows caused by reservoir management would tend on average to taper off as distance increases from the dam, rather than to remain uniform and then end abruptly at 50 km. Furthermore, while communities within 2.5 km of the river would tend to be more affected than those further away, once again the cut-off does not represent an absence of impact beyond that point, nor uniform impacts closer to the river. A potential improvement to the Zone of Influence approach would be to adjust the total size and shape of the zone to each dam based on a potential flood event for instance, the extent of flooding during an ARI event if used to assess flood management during extreme events. However, this would require greater resources to model downstream flooding, and would likely also necessitate a number of assumptions regarding the river characteristics. Upstream impacts are not included in this index for two main reasons. The first is due to a lack of data on the shape of the reservoir, which can differ substantially based on the topography of the area. In contrast, the route of the river downstream of the reservoir is known, even if information on width, depth and flow is lacking. Determining the variation in size of the reservoir due to weather events and management further complicates the problem. Secondly, the nature, severity and extent of upstream impacts on the Population, Agriculture and Environment Indicators may not be comparable to those downstream as the type of flooding differs. Data is also more uncertain. Communities previously located within the reservoir area will have already resettled elsewhere, although they may continue to use the reservoir and its variable area for fisheries and other purposes. Additional information would be necessary on whether new permanent or temporary dwellings or infrastructure have been built adjacent or within the maximum reservoir area Indicator 7 - Population The population score is a proxy for direct human impacts.all else being equal, higher populations within the Zone of Influence will result on average in greater loss of life, injury or other damage to health, including from illness, property damage, and net impacts. The relative impacts on different population groups are not considered, with the exception that for this index poor people are weighted slightly more than the non-poor 8 but does not adjust for differing poverty definitions in each country 9. The poor are more vulnerable to disasters as the marginal utility of their assets is greater, and therefore the loss of an equivalent asset would affect a poor person more. Furthermore, a significant percentage of the poor in the LMB are reliant on subsistence agriculture, and their lack of liquidity places them in greater danger of food insecurity in case of harm to their property or livelihoods. Although the differential impacts of flooding of various groups are not included in the index, it is important to consider how impacts may differ in type and severity. For example, while both in Lao and in Cambodia the great majority of the rural population is engaged in agriculture, a greater portion of the Cambodian rural poor are landless (approximately 12%, as compared to 3% in Lao PDR 10 ). Some of the landless poor are also engaged in agriculture, but as workers or tenant farmers, while others are engaged in short-term services. Damage to agricultural land and food price volatility would impact landowners and the landless differently, even if both are poor. 8 This index weighs the poor 1.5 times greater than the non-poor. Weighting the poor more than non-poor is indicative of the relatively greater impacts flooding would have on the poor than the non-poor in terms of well-being, food security and livelihood. 9 All poverty figures are calculated with respect to national poverty lines. Given higher levels of average income in Thailand, a poverty rate of 40% in a particular Thai district implies a higher level of community welfare than a district with a 40% poverty rate in Cambodia or Lao PDR. 10 See IFAD athttp:// 39

53 Furthermore, the ethnic composition of the population could also result in differentiated impacts. This would partly be due to whether specific livelihoods or livelihood practices associated with an ethnic group are at greater threat from flooding. However, other important issues could influence the severity of the impacts these groups would face and their comparative resilience for instance, access to government services or other forms of support during crisis events may be vary substantially by group 11. Monetized damages are not included in the index. In fact, it is often the case that poverty level is negatively correlated with monetized impacts of extreme events and disasters 12 because the poor have less valuable assets. However, since the Population Indicator weights the impacts on the poor greater than the impacts on the non-poor, this Sub-Index (composed of the three Indicators) would tend to rank the potential impacts of dams in poor areas higher than an index which prioritized monetized damages. Figure 25: Dam locations and Ethnic Areas Downstream impacts of flooding can have differentiated impacts for different population groups, depending on livelihood, gender, age, poverty level, and other characteristics.the Population Indicator, however, does not consider differentiated impacts, except for poverty. Further studies on reservoir management and downstream impacts may consider how these groups are differentially impacted in order to improve management practices. Figure 25 shows the complex ethnic mosaic of the region which in reality is overlapping. Ethnic groups may be associated with particular livelihoods and practices. Some experience higher levels of poverty and may be marginalized from government support during extreme events. Population Parameter The population data was sourced from LandScan Global Population Database 13 (developed at the Department of Energy's Oak Ridge National Laboratory), resulting from a population distribution model that shows a geographical distribution of population at one-kilometre resolution over an average 24 hour period. The algorithm uses spatial data, imagery analysis technologies and census 11 This is due not only to levels of literacy, access to roads, and other measures of development, but also to the relationship of the ethnic group with the local or national authorities. For example, a history of conflict between of some members of ethnic groups to authorities has contributed to the marginalization of these groups and could complicate relief measures in the case of flooding, or good practice management practices, including consultations and communications. 12 This would also likely result in a negative correlation between monetized impacts and the Population Indicator (but not the population parameter), if controlled for population density. 13 See 40

54 data from each country. The population data are estimates that differ slightly from census data, but are considered to be more relevant as they include not only where people live but also where they work. Poverty Parameter Poverty Data was acquired from the respective country government statistics offices. Because poverty data for the LMB countries is only available at the district level, the number of poor people within each Zone of Influence was estimated based on the proportion of poor people in the districts which the Zone of Influence crosses. This calculation implies an even distribution of the poverty ratio within the district i.e. it assumes that for a district with 20% poverty, within the area of the Zone of Influence which intersects with the district, the poverty level is also 20%. However, it is possible that poverty levels in riparian zones are not reflective of the district average poverty levels. First, urban areas in the LMB tend to have lower poverty rates than rural areas. Second, land value within the Zone of Influence may be greater than the value of land further from the river, due to the greater ecosystem services provided. However, it is also uncertain whether this would present a systematic bias toward lower poverty levels, as a significant percentage of the population within the riparian zone are not landowners. This can in fact exacerbate the hardship of those living or practicing their livelihoods near the river, as rental and other costs associated with higher property value may be greater. There is no clear bias for either higher or lower Zone of Influence poverty levels compared to the district average. Main findings The following map shows the geographic distribution of the Population Indicator, with information on poverty rates. 41

55 Figure 26: Geographic Distribution of Dams and the Population Indicator Results The Population Indicator was calculated for 67 of the dams held within the MRC database. The Indicator was found to increase steadily with a low and near constant slope for the 59 dams with lower population levels (in the lowest Population Impact category). The slope then increases dramatically for the remaining 8 dams in the higher Population Impact categories:a small number of dams dominate the population impacts of the group. The results are presented in Figure

56 As show in Figure 27, the highest ranked dams are located in areas with relatively high poverty level, but a large number of dams with a low overall Population Indicator score are located in areas with very high poverty rates (between 31 and 75 % of the population) but in areas with lower population density. Population Indicator scores near 0 were recorded for 14 dams (however, this only indicates low numbers, but not an absence, of people in the Zone of Influence). It should be noted, however, that potential impacts per person for these dams may be quite high due to higher poverty levels, the remoteness of the locations (many of these areas are located in mountainous regions, such as the high poverty corridor along or near the border of Vietnam), and the high percentage of ethnic peoples in this area, who may also be marginalized. The three highest scoring dams Xelabam, Ngam Khan 1 and Nam Ngum 1 also have very low Environment Priority Area Scores, with 0, 0.16 and 0 respectively (Table 10). This indicates that a lower percentage of the area has been identified as a Key Biodiversity Area or as a Protected Area, as expected since in general such areas are associated with high population densities. Figure 27: Population Indicator - Standardised ranking: (0 0.2) 59 dams are in the very low population impact category; ( ) 4 dams are in low category; ( ) 2 dams are in the medium category; ( ) no dams are in the high category; ( ) 2 dams are in the very high category. Table 10 - Population Impact Indicator: Highest scoring dams Name Standardized Pop. Env. Agri. Population Score Rank Rank Rank Xelabam Nam Khan Nam Ngum Buon Kuop Buon Tua Srah

57 3.4.3 Indicator 8 - Agriculture The agricultural indicator is composed of the following ten land-use parameters: Cropland Managed Pasture Cropland/pasture Agriculture with forest Agriculture with other vegetation Agriculture/forest mosaic Agriculture/other mosaic Forest with agriculture Other vegetation with agriculture Agriculture/2 other land cover types Of the land-use parameters described by the MRC, those listed above contribute most to agricultural production and food security. Nevertheless, it is recognized that they are of differing importance in terms of both total food production and with regards to their final consumption, i.e. whether they are used for subsistence farming, or as cash crops for either the local market or for sale to the national or international markets. Furthermore, their relative impact on the four dimensions of food security established by the FAO (availability, affordability, physical access, utilization) 14 is likely to differ substantially. No differential weightings have been applied to these parameters. Other non-agricultural land-use areas may also contribute to food security, sheltering game for hunting, or containing non-timber forest products. Although these areas play an important role in the region for provision of food, especially for lower-income groups particularly vulnerable to food insecurity and during periods of low food availability from other sources, only cultivated land or land used for husbandry was used for this indicator. 14 A set of indicators has been proposed to capture these dimensions of food security however there is little guidance on how these indicators should be combined. See 44

58 Figure 28: Zone of Influence and the Agriculture Indicator Figure 33 shows the Zone of Influence and the Agriculture Indicator which is itself composed of 10 MRC land use data layers each contributing to agriculture. 45

59 Main findings The following map shows the geographic distribution of the Agriculture Indicator and associated dams. Figure 29: Geographic Distribution of Dams and the Agriculture Indicator Results The total agricultural area was calculated for the 67 dams in the MRC database, by combining the areas of 10 mutually exclusive land use parameters pertaining to agriculture and food production, including pasture. As shown in Table 11, the dams with the greatest total downstream agricultural 46

60 area are very strongly associated with a high Population Rank and low Environment Indicator Rank (Protected Areas and Key Biodiversity Areas). A notable exception is the combined Lower Sesan 2 entry, which has both extremely high Agriculture and Environment Indicator Rankings though the positive relationship with the Population Indicator is maintained. At lower levels of the Agriculture Indicator, the positive association with population and negative relationship with KBAs and PAs is less pronounced. As shown in Figure 30, the increase in the Agricultural Indicator is more constant than for the Population Indicator, as well as for the Environment Indicator, indicating a more even distribution of Agriculture across the group of dams than for either the Population and the Environment Indicators (which remain consistently very low for the great majority of dams, and then increase quickly). Nevertheless, the Agricultural Indicator is still dominated by relatively few dams: the higher two categories (high, and very high) encompass only 6 dams, while the lowest two categories encompass 56 dams. Figure 30: Agriculture Indicator - Standardised ranking: (0 0.2) 35 dams are in the very low population impact category; ( )15 dams are in low category; ( )8 dams are in the medium category; ( )1 dam is in the high category; ( ) 4 dams are in the very high category. Table 11 - Agriculture Impact Indicator: Highest scoring dams Name Standardized Agriculture Score Agri. Rank Pop. Rank Env. Rank Xedon Xeset Xelabam Nam Ngum Lower Se San2 + Lower Sre Pok

61 3.4.4 Indicator 9 Environment The Environment Indicator was constructed by joining two non-mutually exclusive parameters: Key Biodiversity Areas (KBAs) and Protected Areas (PAs), and calculating the intersect areas with the Zones of Influence. They may also intersect with areas contributing to the Agricultural Indicator, although the impact that they measure is different and therefore this does not constitute double counting. The purpose of the Environment Indicator is to assess potential impacts of flooding on environmental areas ascertained to be particularly important from an ecosystem service perspective and a priority for protection. Protected Areas and Key Biodiversity Areas also provide ecosystem services increasing foodsecurity. For example, in Cambodia, 75 percent of the rural population depends on access to natural forest resources for products, energy and food, and it is estimated that forest resources account for % of household consumption for 1/3 rd of the population (FAO, 2012). During periods of hardship and low food availability from agriculture sources, these Non-Timber Forest Products (or NTFPs) may account for an even greater percentage of household consumption. However, it is critical to note that even areas that have not been designated as either KBAs or PAs still provide ecosystem services. Furthermore, although KBAs and PAs have been prioritized for protection - and with regard to PAs protected by laws - areas not conferred such status may nonetheless constitute degraded or fragile ecosystems also worthy of protection, if not more so due to the more precarious sustainability of the services that they provide to communities living in their proximity. This Indicator, therefore, places an emphasis on areas already prioritized for protection, and has value in that it not only designates high value environmental areas but it also encourages policy coherence. The caveat, however, is that this Indicator necessarily results in many very low scores, including a great number of scores of 0 - which in part is related to government policy designating Protected Areas, and not only to the ostensible value of the ecosystem. While dams with an Environment Indicator score of 0 may well have less high-value downstream areas, it is more informative to see these dams with a low Environment Indicator score as having a lower environmental policy priority, rightly or wrongly, rather than necessarily lower value. 48

62 Figure 31 Zone of Influence and the Environment Indicator The Lower Sesan 2 Dam ranked very high in terms of the Environment Indicator. Figure 36 shows the intersection of Key Biodiversity Areas/Protected Areas with the Zone of Influence downstream of the dam. For this dam, the population and agriculture indicators were also quitehigh compared to other dams which ranked highly for environment, giving it a high overall Impact Potential subindex score, and ranking 49

63 Main findings The following map shows the geographic distribution of the Environment Indicator and associated dams. Figure 32: Geographic Distribution of Dams and the Environment Indicator Results There is a clear concentration of dams which score very highly with regard to the Environment Indicator (implying a greater potential threat to Protected Areas or Key Biodiversity Indicators) in or 50

64 near the Sesan sub-basin (including two dams in southern Sekong), and to a less extent the Nam Ngum and Sre Pok sub-basins. As expected for this indicator, a very large number of dams have a score of or very near zero 30 dams, almost half of the total. The distribution of the Indicator is therefore very skewed. However, if zero scores are excluded, the Environment Indicator displays the most even balance between categories. Further analysis could investigate whether there is a correlation between dams with greater Environment Indicator scores and their age, since newer dams may be increasingly located in remote areas as the more accessible locations have been developed already, or in order to avoid populated areas and potential stakeholder impacts and resistance. Eight dams are in the high and very high categories. However, for the highest scoring dams, the relationship between the Environment Indicator and the other indicators is not striking 2 of the 5 highest scoring dams also have Population Rankings in the top ten while the other 3 are in much less sparsely populated areas, while 3 of the 5 have Agricultural Rankings in the top ten. Figure 33: Environment Indicator - Standardised ranking: (0 0.2) 42 dams are in the very low population impact category; ( )11 dams are in low category; ( )3 dams are in the medium category; ( )6 dams are in the high category; ( ) 2 dams are in the very high category. Name Table 12 - Environment Indicator: Highest scoring dams Standardized Environment Score Env. Rank Pop. Rank Agri. Rank Nam Kong Lower Se San2 + Lower Sre Pok Xe Pon Nam Khan Plei Krong

65 3.4.5 Conclusions The flood impact potential is a measure of the people and their agro-ecological assets within the riparian corridor immediately downstream of each reservoir. A cumulative score was assigned to each dam based on three indicators designed to assess the FloodImpactPotential downstream of each dam.the indicators utilised included: (i) population indicator, (ii) agricultural indicator, (iii) ecological indicator. The Population Indicator combines poverty level data with the population data 15 to give a weighted population score; the Agriculture Indicator consists of ten land use parameters; and the Ecological Priority Indicator consists of Key Biodiversity Areas (KBAs) with Protected Areas. In general, the Population Indicator and the Agriculture Indicator are correlated, while both are negatively correlated with the Environment Indicator. The results of top and bottom ten ranked dams are shown in Table 13 and Table 14 below. Table 13: Highest relative FloodImpactPotential: Reservoirs with the greatest potential to cause downstream damage RANK NAME 1 Xe Labam 2 Lower Sesan 2 3 Nam Khan 1 4 Plei Krong 5 Xe Pon 3 6 Xe Don 2 7 Nam Ngum 1 8 XeSet 1 9 Buon Tua Srah 10 Nam Kong 2 Table 14: Lowest relative FloodImpactPotential: Reservoirs with the least potential to cause downstream damage RANK NAME 58 Nam Mouan 59 Xe Kaman 2B 60 Nam Mang 3 61 Se San 3 62 Nam Pot 63 Nam Leuk 64 Xe Katam 65 Nam Chian 66 Nam Theun 2 67 Nam Ngiep Reservoir Flood Risk Index: Combined Downstream Flood Potential and Flood Impact potential ranking The Downstream Flood Potential and FloodImpactPotential results were combined to create a Reservoir Flood Risk Index. A summary matrix of all dams considered and their relative scores is presented in Figure 23 below.table 15 below shows dams identified by the Index as priorities for action due to their very high Downstream Flood Potential and very high Flood Impact Potential. There was one dam ranked as very high flood risk (Lower Sesan 2) and two with high flood risk (Nam Khan 1 and Xelabam). Both Nam Khan 1 and Xelabam (upstream of Pakse) have a very high downstream population as they are close to the mainstream. Xelabam has a low level of flood control, however, Nam Khan 1 was ranked well into the upper half of the dams. Lower Sesan 2 Dam also has a 15 Producing two parameters country defined poor and non-poor totals. 52

66 sizeable population downstream and is located at the confluence of two major river systems (Sesan and Srepok), giving it a very high Natural Flood Threat. Figure 34: Reservoir Flood Risk Index scores for Mekong hydro-electric reservoirs Table 15: Reservoir Flood Risk Index priorities: Very high and high Downstream Flood Potential and Impact Potential FLOOD RISK NAME Very High Lower Sesan 2 High Nam Khan 1 High Xelabam 53

67 4 Basin wide priorities for integrating flood control into Mekong hydropower development The matrices produced in the development of the Reservoir Flood Risk Index can be used to identify basin wide priorities for integrating flood control into Mekong hydropower development (Figure 35). The Reservoir Flood Index matrix identifies which dams have the highest flood riskin terms of flood occurrence and impact if a flood occurs. Dams ranked high or very high in this matrix are the priority dams for action. The Downstream Flood Potential Index matrix helps to identity and prioritise management and designoptions to reduce the Downstream Flood Potential of a reservoir. The matrix can assist reservoir managers to understand if a dam needs structural design changes to improve flood control capacity or whether the design is sufficient and efforts should be focussed on improved management. Figure 35: Using the Reservoir Flood Index and Downstream Flood Potential Index to prioritise reservoirs and management responses 4.1 Responding to the flood index The Reservoir Flood Risk Index identifies which dams are priorities for intervention based on the Downstream Flood Potential and the Flood Impact Potential. Analysis of the scores together with the main drivers contributing to high or lowdownstreamflood Potential and Flood Impact Potential can be used to aggregate the projects into comparable flood risk groupings to help set priorities for intervention (Figure 36): Group A first priority:dams with high/very high Downstream Flood Potential and high/very high Flood Impact Potential should be singled out for prioritised intervention Group B1 equal second priority: Dams with high/very high Flood Impact Potential are a priority for action due to the likely high impact on downstream populations if a flood occurs Group B2 equal second priority: Dams with high/very high Downstream Flood Potential are a priority for action due to the higher likelihood of floods occurring 54

68 Group B3 third priority:dams with high Downstream Flood Potential but only medium Flood Impact Potential are a priority for action because it is likely that downstream floods will occur and the impacts may be relatively major Group C fourth priority: Dams in Group C are a low priority because downstream floods are less likely to occur and if they do there is less potential for impacts downstream Three reservoirs are singled out as priorities for intervention (Group A) including Lower Sesan 2, Nam Khan 1 and Xelabam.Most reservoirs are located in the medium to low priority areas delineated by B1, B2, B3 and C. Of these, the initial priorities should be B1 and B2 because of the higher likelihood of floods or the higher potential impact if a flood occurs. Whilst it is still important to ensure good flood control of dams in Group B3 and C, the urgency for action is lower compared to dams with higher scores. Figure 36: Developing basin wide priorities for reservoir action on flood control: A indicates dams at high and very high risk where action should be; B1 indicates dams where improved reservoir design is the priority and C indicates areas with low flood threat where flood control services are likely not required. C B2 B3 B1 A Basin wide priority for action The Lower Sesan 2 + Lower Srepok 2 Dam Lower Sesan 2 Dam has the highest Reservoir Flood Risk Index rating due to its poor capacity to control floods, high natural flood potential and sizeable downstream population at risk.lower Sesan 2Damis a planned400 Mega Watt (MW) hydropower reservoirlocated in Cambodia, just past the confluence of the Sesan and Srepok rivers and downstream of two large cascades (Figure 37). According to the MRC database, it is has a height of 45m 16 and the greatest maximum storage area of the 67 dams assessed. It also has the largest contributing catchment area (two river basins) and subsequently, the highest 100 year ARI and PMF inflows.the project has been heavily criticised for its potential environmental and social implications tofish biodiversity and livelihoods.many people will also require resettlement. 16 According to MRC database, it is 45m although other available reports place it at 75m. 55

69 Figure 37: Location of Lower Sesan 2 Dam: approximately 1.5km downstream of confluence of Sesan and Srepok rivers; 25km upstream of the Mekong River. Shows large number of existing and planned upstream dams, and a downstream population at risk that includes major township of Stung Treng and inhabited areas along Mekong mainstream. The existing design of Lower Sesan 2 Dam will not be sufficient for the project to act as a reregulating hydropower dam mitigating downstream flood risks 17.The dam is at the bottom of two major basins with many existing and planned dams located upstream that would have larger storages to impound flow. This implies that the dam is designed for re-regulation, however, in practice it may not function as such. Experience from other regulating dams on the Sesan suggests it would be solely operated for electricity production, thereby causing water fluctuations downstream.lower Sesan 2 dam s location would also require it to have a larger spillway capacity to pass through releases from the other dams if they rapidly dischargeduring an extreme flood event. Results contradict this however, showing that thelower Sesan 2 is in the lower half of the 67 dams analysed when 17 Controls unnatural fluctuations in water level and flow during peaking operations of upstream dams and releases it to match natural conditions. 56

70 comparing the discharge capacity to the 100 year ARI peak flow rate 18.This would be a minimum required discharge rate given the location of the dam. As the dam is still in the design phase, consideration should be given to increasethe capacity, although it will depend on the design of the spillway. The planned reservoir drawdown is only 1.0 meter, which is why the live storage volume was rated quite low compared to the mean annual flow, resulting in a reduced regulating capacity. In practice however, the storage of dams in the upper reaches of the rivers would impound the majority of mean annual flow and Lower Sesan 2 Dam slarge storage area will provide substantialflood routing.given this large surface area and the surrounding flat topography, raising the dam to include flood bufferstorage is not recommended as any increases would further inundatehuge areas ofland and add to the already considerable environmental and social issues associated with the project. Another option could be to lower the reservoir s full supply level to allow dedicated flood storage, which is potentially feasible as the spillway design proposed includes gates, but would need further investigation. The transboundary nature of the Sesan and Srepok basins compounds the issues ofrelatively small spillway discharge and regulating capacity. The rivers begin in Vietnam and end in Cambodia, which brings with it differing standards in operating and management protocols, and is a likely hindrance to communications between dams and countries for releases. This potential lack of coordination during a flood event could lead to an unexpected inflow much greater than Lower Sesan 2 can discharge safely, or at all.poor communication has resulted in severe impacts in the past, wheredischarges from Yali Falls Dam have caused flash floods that have killed people in Cambodia.There is a sizeable population downstream of the dam, including the major town of Stung Treng (25km downstream), whichmust be protected from large releases. Measures for mitigating flood risk for this dam should focus on clearly defined and stringently enforced management not only for Lower Sesan 2, but also across the basins. It is important to ensure that upstream dams contribute to flood control and communicate effectively across a cascade and between countries when releasing water. This will avoid putting the entire onus to control floods from Sesan and Srepok rivers onlower Sesan 2 Dam. A priority design measure for this dam would be to: Improve the spillway discharge capacity by widening and deepening the opening or installation of an auxiliary spillway. The main management and regulation measures for the dam (and the two basins) is as follows: Standardised flood forecasting and early warning system proceduresto facilitaterelease coordination between countries, cascades and dams that will limit potential damages to populations at risk. Emergency protocols for dam operationthroughout the two basins including dam discharge rule curves and clear instructions on required actions and responsibilities during an extreme flood event. Regulatingof dam operations by an independent authoritythat will ensure operators are well trained and following the required guidelines and standards for flood control. 4.2 Responding to thedownstream Flood Potential Index The Reservoir Flood Risk Index highlighted which dams are priorities for intervention, whereas the Downstream Flood Potential Index will allow identification of management and design options to improve flood control capacity of these priority dams. The basin-wide priorities are shown in Figure 38. The largest portion of reservoirs have very low or low Flood Control capacity and are located in areas of low Natural Flood Threat. Whilst it is still important to incorporate flood considerations into the 18 The spillway has gates and a capacity that is 1.66 times the size of the calculated 100 year ARI peak inflow. 57

71 design and management of these reservoirs, the urgency is lower compared to dams with higher scores. Analysis of the scores together with the main drivers contributing to high or low threat and control capacity can be used to aggregate the projects into comparable flood risk groupings which is useful in terms of setting potential response measures (Figure 38): Group A: Dams in areas of high Natural Flood Threat and with high Flood Control capacity can optimise their flood control benefits through revising reservoir management. Group B:Dams that are located in areas of high Natural Flood Threat and lowflood Control capacity need to be prioritised for design changes to increase their control capacity. Group C:Dams in very low Natural Flood Threat areas do not need to prioritise Flood Control in their design or operation. Figure 38: Developing basin wide priorities for reservoir flood control:a indicates dams where improving reservoir management is the priority; B indicates dams where improved reservoir design is the priority and C indicates areas with low flood threat where flood control services are likely not required. C B A High control capacity dams in areas of high flood threat Group A Dams that have high potential for Flood Control and are located in catchments of high Natural Flood Threat may not currently include provisions for flood storage, but have the capacity for management reform to do so. From the analysis, it can be seen that no dams were found to be in the Group A category, which may be a result of insufficient design standards followed in the Basin or because not all existing and proposed dams were included in the study. If the study was taken further to include all large dams and found that some were classified into the Group A category, potential management reforms to incorporate flood control could include: (vi) Flood forecasting through hydrological monitoringto inform dam discharge operation. This could help to draw down the storage before a flood event. It should be noted that this is only an aid for decision-making. (vii) Early warning systems for downstream inhabitants to notify them on dam releases (whether the releases relate to normal operations or flood events). 58

72 (viii) Installation of gated spillways to provide greater flexibility in controlling discharges, although gated spillways do not guarantee flood mitigation. (ix) Lowering of the dam design full supply level to incorporate dedicated flood buffer storage. This storage would remain empty unless it is required to store floodwater. This would be discharged slowly to reduce the peak flow downstream. (x) Dam discharge rules that include consideration of timing of downstream tributary peak flows to avoid coinciding peak flows. Many dams in the basin could have high potential to manage downstream flooding with minor alterations in their operating regime. In addition, from a flood control point of view a number of existing dams (Nam Ngum 1 and Xe Bang Nouan) have a high potential within their existing design and would be well suited for use as a regional demonstration project in fully integrating flood control into reservoir operations and management. Nam Ngum 1dam in particular is part of a larger hydropower cascade which increases the complexity of management but also the downstream damage potential from poorly managed flood flows Low control capacity dams in areas of high flood threat Group B A larger number of dams are in high Natural Flood Threat areas but have limited capacity to control flood flows. Thereis significant potential to improve flood control through changes to design of the projects and their components. Design reform to include this should take place at the outset of project pre-feasibility assessments and concessions should be determined before firm agreements on the quantum of electricity production are negotiated so that these projects move forward as true multiple use dams. Inclusion of flood control design measures should focus on the reservoirs identified as medium to very high NaturalFlood Threat and very low Flood Control capacity (Table 16). Of the dams categorised as Group B, the following have been found as priority based on the Reservoir Flood Risk Index results: First prioritybecause of high Downstream Flood Potential and Flood Impact potential:lower Sesan 2 and Xelabam Second prioritybecause of: o o High Flood Impact potential: Xe Kong 3D High Downstream Flood Potential and very low to medium Flood Impact potential: Xe Kaman 2A, Xe Kaman 2B, Dak E Mule, Xe Kong 5, Xe Kaman 4A, O Chum 2, Xe Kaman 4B, Xe Kaman 3, Xe Kong 3up, Xe Katam, Nam Kong 2, Nam Phak and Houay Lamphan The planned projects in this list should be the focus of efforts to support developers to revise designs for enhanced flood control. Table 16: Reservoirs with very low flood control capacity located in areas of medium high NaturalFlood Threat: planned projects are underlined and bold NAMES OF RESERVOIRS NATURAL FLOOD THREAT FLOOD CONTROL DOMINANT INDICATOR LIMITING FLOOD CONTROL Lower Sesan 2 Very high Very low Xe Kong 3D Very high Very low Limited live storage compared to the mean annual flow; limited spillway discharge capacity Limited live storage compared to the mean annual flow Xe Kaman 2A and Xelabam Medium Very low Limited live storage compared to the mean annual flow; and limited emergency storage volume O Chum 2 Medium Very low Limited emergency storage volume; limited spillway discharge capacity 59

73 NAMES OF RESERVOIRS NATURAL FLOOD THREAT FLOOD CONTROL DOMINANT INDICATOR LIMITING FLOOD CONTROL Xe Kaman 4A, Xe Kaman 3, Xe Kaman 4B andxe Kong 5 Medium Very low Limited emergency storage volume Xe Kong 3up, and Xe Kaman 2B Medium Very low Limited live storage compared to the mean annual flow Dak E Mule, Nam Kong 2, Nam Phak, Xe Katamand Houay Lamphan Medium Very low Dam discharge capacity The reservoirs located in medium to high NaturalFlood Threat areas havevery low flood control capacity as they were deficient in one or more of the three criteria of the Index. Six projects had reduced Flood Control due to limited livestorage compared to inflows, whereas five projects had poor emergency storage volumes. Five dams were ranked very low for Flood Control due to the dam discharge capacity. The redesign for dams located in medium to high Natural Flood Threat areas should focus onimproving the emergency storage volume by increasing the freeboard or specifically designing and operating for a flood storage volume. Other options exist such as the increase in the discharge capacity of dam spillways, however such measures should be considered in the full context of the potential environmental and social implications of increasing capacity for large releases Dams in areas of low flood threat Group C Dams that are located in catchments with low NaturalFlood Threat need not necessarily include specific measures for flood control. Thesewould be the best suited of the projects considered to continue operation solely for electricity production without flood control (Table 17). Table 17: Reservoirs with low flood threat: Nam Ngum 1 and 2 are existing dams, whereas Nam Suang 1, and Nam Khan 1 & 2 are planned. DAM NAME NATURAL FLOOD THREAT FLOOD CONTROL DOMINANT INDICATOR PROVIDING LOW FLOOD THREAT Nam Khan 1 Very Low Very Low Low Probable Maximum Flood Nam Khan 2 Very Low Low Low Probable Maximum Flood Nam Suang 1 Very Low Very Low Low Probable Maximum Flood Nam Ngum 1 Very Low Medium Low Probable Maximum Flood Nam Ngum 2 Very Low Low Low Probable Maximum Flood The key indicator that provided these dams with a low NaturalFlood Threat was the low estimated Probable Maximum Flood. For each of these dams, the 100 year ARI was also found to be relatively low. This low flood threat is likely due to the dams being in larger catchments with relatively low precipitation compared to other regions of the basin Dams in areas with short flood response times Reservoirs with times of concentration of less than six hours mean that operators and downstream communities have limited time to respond once a large storm event begins. It is therefore important that these dams have good flood storage capacity to create a time buffer for flood preparations downstream and to begin controlled releases. These dams should also have the ability to provide controlled release of large flows, which willrequire gated spillways with a highdischarge capacity. Table 18presents the reservoirs found to have short times of concentration and poor capacity to store and discharge flood flows. The twelve of these dams that are not yet commissioned should be the focus of structural redesign and improved operational procedures to ensure better consideration of the quick response times of the catchments. Redesign and improved operational procedures may include: 60

74 (v) Installation of gated spillways to allow improved control of discharge; (vi) Increasing the dam discharge capacity so that the dams can quickly release water on short notice; (vii) Creating a time buffer for downstream communities by specifically designing and operating for a flood storage volume; and (viii) Installing improved flood forecasting and early warning systems in parallel to construction of the reservoir. Table 18: Reservoirs with time of concentration less than six hours and low capacity to store and/or discharge flood flows TIME OF NAMES OF COMMISSIONED CONCENTRATION RESERVIOIRS NAMES OF PLANNED RESERVOIRS 0-2 hours N/A Nam Hinboun 2, Nam Pay and Nam Phak 2 4 hours O Chum 2 Nam San 3B, Dak E Mule, Xe Kaman 4A and Xe Kaman 4B 4 6 hours N/A Nam Chian, Xe Lanong 2, Xe Set 3 and Nam Mang 1 61

75 SECTION C SUPPORTING MATERIALS AND ANNEXES 62

76 5 Supporting notes on incorporating flood control for Mekong hydropower dams Hydropower dams located in flood prone areas of the Mekong Basin need to prioritise incorporation of flood control measures.some of these measures may already be in place and operating effectively, however, in many cases they are not due to limited oversight and the associated political economy. The purpose of this section is to briefly introduce internationalbest practice in flood control, with design, management and regulationoptions that could be adopted in this region. 5.1 Understanding the limitations Before consideringhow to integrate flood control into dam design and operation, it is important to have realistic expectations of what is possible, as even the largest reservoirs cannot control the most extreme floods. With increasing flood return intervals, a dam s ability to mitigate peak flows downstream decreases due to the larger volumes of water (Figure 39). Reservoirs can only attenuate floods up to a certain size after which it must be passed directly through the spillway to avoid dam failure. Figure 39: Effect of reservoirs on the flood peak downstream:mitigation is limited the higher the flood return period(source: Flood control is really only possible in reservoirs that are large compared to the size of the river that feeds them. If the reservoir is too small, the water simply fills the reservoir and runs through, without much change in the magnitude of the flood. The operator of the dam has no choice but to release the flood water at the same rate as the inflow. Therefore, flood control should be targeted on(i) planned dams to bring in design changes that ensure the reservoir is capable of providing the capacity, and(ii) existing dams in flood prone areas that have the physical capacity to provide control, are able to be retrofitted and can be managed differently. Dams that have been successfully controlling floods over decades can also increase potential downstream impacts in the event of extreme floods becauseof community behaviours. Once a dam has been constructed and is successful in regulating smaller floods, downstream inhabitants become used to the new flow regime. Believing the dam can protect them against all flood events, they encroach on floodplains and relocate to areas along the riverside that may not have been safe prior to dam construction. When an extreme flood occurs that the dam cannot sufficiently mitigate, the community is then severely impacted and damages are higher than they would normally be without a dam. 63

77 In summary, large dams are generally not an effective method of controlling floods due to the issues raised above as well as(world Commission on Dams, 2000): The high costsassociated withensuring protection against extreme floods, Theirdiminishing efficiencyover time due to accumulation of sediments in the reservoir, and The reduction and elimination of natural flooding that has affected important ecosystem functions and livelihoods for flood dependant communities. Given however that hydropower development is likely to go ahead in the LMB regardless of these concerns, incorporation of flood control measures should be a priority.understanding these limitations will inform development of a more robust set of measures that will havea greater chance of success at maximising the benefits of flood control that dams can provide. 5.2 Methods of flood control There are three main groups of flood control measures, being (i) design measures, (ii) management responses, and (iii) policy & regulation. These are described in more detail in the following sections Design measures Design measures involve altering the design of the physical infrastructure of a hydropower dam to more effectively store and discharge flood water. These responses are best undertaken at the planning and project design phases, as retrofitting a dam can prove costly. Appropriate siting Appropriate siting of the dam in the catchment andrelative to the location of communities is important, as this determines the extent to which it can reduce downstream flooding (Box 3). Siting of a dam can only be decided on during the initial stages of project development, and is therefore a critical parameter. A dam that is located in the upper reaches or on a minor tributary within the catchmentwill impound only a portion of runoff andtherefore provide limited flood control benefit for the populationbeyond immediate downstream communities. Any flooding from tributaries and/or heavy rainfall downstream of the dam are beyond its ability to influence.for the same reason, a dam that is intended to provide flood control for downstream communities but is located far upstream is not likely to effectively control floods. Siting of the dam should consider these factors to equally enhance flood control and maximize electricity generation potential. Box 3: The percentage of a catchment commanded by a dam heavily influences its flood control the case of Wivenhoe Dam in Australia illustrates limited influence on flooding in Brisbane during a major event in Wivenhoe Dam is a water supply and flood mitigation dam in the Brisbane River Basin, upstream of the heavily populated city of Brisbane, Australia. It controls approximately half of the total Brisbane River Catchment Area, but does not control the catchments of Lockyer Creek and Bremer River. Releases from Wivenhoe dam can however be timed to avoid coinciding peaks with these tributaries. Heavy rainfall in produced major floods in Brisbane, which were initially blamed on Wivenhoe Dam operations. Widely varying rainfall forecasts and unpredictable heavy localised rainfall caused delayed dam releases which then coincided with peak flood flows from the other tributaries, exacerbating flooding. These floods did not however originate solely from the dam sub-catchment. The dam actually provided a flood peak reduction of 40%, equating to 2.0m decrease in Brisbane River water levels in Brisbane CBD, but could not eliminate flooding altogether as it did not control the entire catchment. 64 The experience of Wivenhoe is an important reminder of the need for proper understanding of the sources of downstream flooding and the significance of siting in determining the effectiveness of flood control.

78 Increasing emergency storage volume Increasing the available emergency storage volume, which is typically between the FSL and dam or spillway crest level, will provide additional time for operators to respond to large and unexpected flood inflows. This storage volume can be augmented by (i) raising the dam wall, (ii) altering the spillway arrangement, or (iii) lowering the FSL by changing the operating rules (a design and management measure). The first two options are costly, as they involve structural changes to the dam although they will not result in lost generation revenue. The third option however will require a tradeoff between flood control and electricity production that may be justifiable from an economic standpoint because of avoided flood damage costs. Incorporation of flood buffer storage Allowance for flood buffer storage will ensure that dams always have volume available for attenuation of flood events. Flood buffer storage is a dedicated space above the FSL (i.e. live storage) of the dam that is used to temporarily store and discharge flood waters. This storage space is never utilised during normal operations, and water must be discharged as soon as possible to ensure the space is free to store future flood water. There are two types of flood buffer storage, being (i) permanent, and (ii) seasonal. In catchments where floods can be expected at any point in the year, permanent storage capacity should be kept free to impound flood water. Seasonal allocation can be more appropriate in catchments where historical analysis of floods has shown a distinct time period (i.e. wet or cyclone season) where floods occur, although this should take into account the potential impact of climate change on weather patterns. (Sinha & Srivastava, 2011) A planned dam may be designed to be larger than it normally would be in order to incorporate the buffer storage and also maintain the required electricity output. This however will require greater cost and the larger inundation area could have potential environmental and social implications. Increasing spillway discharge capacity The spillway discharge capacity can be increased to allow dam operators to rapidly releasewater and lower the reservoir level under flood conditions. This could be done by improving the capacity of the existing spillway structure or construction of an additional auxiliary spillway. Both options will require significant financial outlay. It is important to note that this will protect dam safety but does not necessarily result in improvements to downstream flooding, as the rate and timing of discharge must be carefully regulated with operating rules to avoid worsening flood conditions (e.g. coincident flood peaks) downstream of the dam. Installation of gated spillways Gated spillways (Box 2) will allow earlier and greater control over the dam water levels and the rate of discharge. If located lower in the dam wall (i.e. well below FSL), the gated spillway can be used to prerelease the early stages of a flood. An important factor to consider with gated spillways is possible mechanical or electrical failure in an emergency situation and resultant dam safety issues, for which there has been a precedent.having well trained operators and a rigorous maintenance program for the mechanical and electrical equipment, including frequent dam safety inspections and dry runs of emergency scenarios, will prevent these issues. Analysis of the MRC database shows that 67% of the large hydropower dams already have gated spillways installed.conversion of uncontrolled to gated spillways is a costly exercise but could be the difference between effective and non-effective flood control for a dam. It is important to note that installation of gated spillways does not guarantee flood control, as it will depend on the operation of the infrastructure Management measures Flood control management measures alter the operating practices of a hydropower dam to more effectively control the storage and discharge of the water.this should be implemented for existing and planned dams. 65

79 Flood forecasting Hydrological monitoring and flood forecasting (short and long term) can be used to predict floods before they occur, thereby feeding information into early warning systems for local people and informing dam operations early enough to allow rapid but safe discharge of water prior to the flood wave arriving. The aim is to reduce the flood peak whilst minimising any downstream damages. Flood forecasting includes the(pagasa, 2014): Installation of telemetry systems to collect hydrometeorological data on a real-time basis, including information obtained from rainfall and water level gauging stations, Assessment of weather conditions affecting or expected to affect the reservoir catchment area, and Use of a real-time processing computer system to undertake streamflow simulation and forecasting of flood inflow into the reservoir from telemetered rainfall data. The MRC provides one to five day flood forecasts for 23 locations along the Mekong River (Pengel et al, 2008), however, these systems are not widely used in the tributaries where the bulk of existing and proposed dams are located 19. Availability of information such as this, along with appropriate training and regulations will ensure that the dam operators make the most informed decisions. It is however important to note that such a system is only an aid to hydropower operators and cannot be used as a sole basis for decision-making. Early warning systems Developed on the basis of flood forecasting, early warning systemsarecritical for emergency services to evacuatecommunities in the path of a flood. In addition,the more lead-time local people have, the better placed they are to preserve their valuables, which reduces damages. The typical components of an early warning system are shown in Figure 40, and include (i) flood forecasting, (ii) warning, and (iii) response. Figure 40: The components of an early warning system: (Krzysztofowicz & Davis, 1983) The flood forecast is calculated and transmitted to the decision maker, who prepares a warning based on their assessment of the situation.the warning is then disseminated to relevant local authorities and communities using multiple communication mediums (phone call, SMS, facsimile, television, radio, internet, loudspeaker, billboards) who then act according to the relevant emergency procedures. (Plateand Insisiengmay, unknown) The critical path in this process is relaying the warning effectively to local communities as emergency services will likely not be able to reach all areas. As mentioned previously, the MRC does operate an early warning system that focuses on the mainstream, andcertain areas may have detailed procedures in place, such as Srepok Basin (see ICEM and IWRP, 2014). These may not always be effective as evidenced in recent years by news reports of 19 The MRC is also developing a flash flooding guidance system for tributary rivers which will indicate the likelihood of flooding of small streams over wide areas ( 66

80 dam releases exacerbating flooding downstream (e.g. A Vuong Dam, although not in the LMB, is a typical example - Box 1). Incorporate flood buffer storage Incorporating flood buffer storage into reservoir operations of an existing dam is a management measure that will result in a reduction in live storage and therefore power generation potential, whereas a planned dam can be redesigned. This is a relatively simple method of including flood control that can be economically justified against lost electricity potential on the basis of avoided downstream environmental and social damages. There are regulations in parts of the LMB (e.g. Srepok Basin Viet Nam portion, see ICEM and IWRP, 2014) stipulating that reservoirs have to use part of their storages for flood protection during the flood season, however, this is not enforced adequately in all areas and is not in place throughout the LMB. Dam discharge rules Also known as reservoir rule curves, dam discharge rules control the amount and timing of releases from a dam(figure 41).These rules are formed from data on: (i) required water allocations (irrigation, water supply, flow augmentation/environmental flows, electricity generation), (ii) the types of infrastructure and number of inhabitants downstream, (iii) a history of past extreme storms in the catchment, (iv) the capacity of the downstream channel and (v) inflows (Mistry, 2011). Dams can have more than one rule curve that is used for different purposes depending on the situation and its priority. These curves are presented as minimum water levels or discharge rates and inform dam operators on how to manage the reservoir to meet the requirements of the situation. Figure 41: Examples of dam discharge rule curves LHS:taken from Belwood Reservoir on the Grand River, California (GRCA, 2013), RHS: example of multiple use curves (Mistry, 2011) The rules should also direct operators on how to maximise the availability of flood storage and most effectively regulate flood peaks downstream. It is important for dam operators to have access and adhere to a well-conceived rule curve during the flood season that is periodically updated to reflect improved knowledge of the basin and long term shifts from climate change (ICEM, 2012b). This will provide operators with a clear set of instructions to ensure dam safety during an extreme event and avoid panicked and sudden release of flows that may cause damage downstream Policy & regulation The design and management measures outlined in the previous sections are international best practice and if implemented would address shortcomings in flood controlfor reservoirs in the Basin. Some of these measures may already be in place on certain dams. There is however a gap that is yet to be bridged, which is how these actions can be regulated, enforced and standardised across the 67

81 Basin. Flooding is a transboundary issue and cannot be dealt with differently in each region. Development of appropriate policy and regulation would close this gap and ensure that multiple use for flood control is brought into the regional context. Spatial zoning Spatial zoning should be considered from a regional scale as well as on a project scale. In the regional context, spatial zoning should be undertaken to identify and map areas with high flood threat. If a hydropower project is already operating or being proposed in such an area, government regulations should firstly ensure that it is safe to construct there, and then make it a priority to incorporate relevant flood control measures into the design and management of the reservoir. On a projectscale, flood inundation maps should be provided to urban planners and zoned into high and low threats so that flood prone areas are utilised for non-critical purposes, such as recreational activities, and important household and business functions can be moved to higher elevations. Emergency protocols for dam operation Emergency protocols could be incorporated in (or referred to as) a disaster management plan. They describe a set of procedures for operation of the dam under flood conditions, andare also likely to include instructions for other types of emergencies (earthquake, fires, failure of fish facilities, generation deficit). Emergency protocols for flood control includeinformation such as: Pre-emptive actions to diffuse the beginnings of the emergency, Emergency response if pre-emptive actions are not successful, Outline of responsibilities, such as who will be the decision maker for dam operation in an emergency - in some cases the regulating agency will take over operation, or it could remain with the dam operator - andshould also describe which staff should do what, Discharge rule curves, Who to notify when discharging and how to do so, and What to do when communications or equipment fail. Guidelines and standards It is critical that there are uniform guidelines and standards for hydropower design, construction, operation and management in place throughout the Basin to ensure that there is an established understanding between dam owners, operating agencies, and government that will reduce potentially controllable flooding incidents. These documents should be clearly defined and include minimum requirements for size and operation of a flood buffer storage and the spillway (including the influence of climate change), disaster management plans (including flood forecasting, early warning, provision of inundation maps, dam discharge rules, evacuation strategies), and coordination within cascades. They should also stipulate that local government/authorities develop a more detailed set of instructions that reflect local conditions. Regulating authorities Regulating authorities and legal frameworks ensure that dam operators are following the relevantguidelines and standards for flood control.according to the World Commission on Dams (2000), compliance usually fails when standards and regulatory frameworks lack legal requirements and are weak. In addition, it found that donors and governments tend to ignore them as they may be incomplete or incoherent, ill defined and have low levels of monitoring and accountability. The regulating authorities shouldbe given adequate funding and staffing to fulfill their responsibilities. It is also importantfor the authority to have the power to punish offenders with sizeable fines to ensure that there is deterrence to breaching the rules. Legal framework should be put in place to support this. 5.3 Entry points for change - the hydropower planning cycle Flood control is marginalised largely because of the way hydropower planning is implemented in the Basin. The development cycle is undertaken almost exclusively within the electricity sector and therefore results in dams that are optimised for sole use electricity generation. In order to bring flood 68

82 control back into the conversation, the measures outlined in the preceding section should be incorporated into the hydropower planning cycle. Figure 42 shows the typical lifecycle of a hydropower project, from development of a strategy through to detailed design and ending with refurbishment or decommissioning. Figure 42: The typical hydropower planning cycle Strategy:development of an orientation strategy for the hydropower sector including identifying the needs to be met, such as the required generating capacity Master Plan:plan produced by the government that outlines the direction of hydropower development, and could include a listing of all possible projects Plan: specific plan for a region/basin, which may involveinitial scoping and reconnaissance Project Design: includes the pre-feasibility, feasibility, and detailed design of the hydropower scheme or dam Project Implementation:the procurement process is undertaken before the design is constructed and then commissioning, ready for operations to begin. Refurbishment (Decommissioning): end of the cycle and stage at which the plant is either refurbished to extend its operational lifespan or decommissioned. PR OJ EC T IM PL E M EN TA TI O N Certain groups of measures are more effective if they are incorporated at specific points in this cycle (Figure 43), such as: Management responses:should be included in the project implementation step. This would be for existing dams that are already operating or are being refurbished and transferred, have the potential capacity to reduce floods and are not already operated to do so.dams with a high Natural Flood Threat and high Flood Controlcan benefit from this, as can dams with high Natural Flood Threat and low Flood Control although this would be to a lesser extent. Design responses: should be input at the project plan and design steps. This is for planned dams where flood control can be incorporated into existing designs and intended management of the scheme. Dams with a high Natural Flood Threat and low Flood Control Potential can benefit from this. It is also relevant for existing dams that can be retrofitted or are being refurbished. Policy and regulation: should be implemented at the strategy, master plan and plan stages. This positively influences the flood control performance of all existing and planned dams. PROJEC T PLANNI NG PHASES 69

83 Figure 43: Entry points to the hydropower planning cycle-cycle restarts after Refurbishment or decommissioning, where there is an opportunity to Build, Operate, Transfer the infrastructure 70