A comparison of hydrological models for forest management and climate change

Transcription

1 A comparison of hydrological models for forest management and climate change How can existing models capture and reproduce hydrological effects of forests? Authors: M. Adams, G. Markart and U. Stary Department of Natural Hazards and Alpine Timberline Federal Research and Training Centre for Forests, Natural Hazards and Landscaper (BFW) Rennweg 1 Hofburg A-6020 Innsbruck CC-WaterS is supported by means of the European Regional Development Fund (ERDF) and by the Instrument for Pre-Accession Assistance (IPA)

2 2 Table of Contents Abstract Introduction Overview of the types of hydrological models Black-box models Deterministic / physical models Conceptual models Lumped models Distributed models Semi-distributed models Aspects concerning land-use changes in hydrological models Model reviews in the current literature Comparison of models WMS models HEC 1 (HEC-HMS) TR MODRAT Rational Method GSSHA Other models ZEMOKOST TOPMODEL HYDR²AC BROOK Hydrologic-Models CATFLOW PREVAH WaSiM-ETH...38 CC-WaterS Climate Change and Impacts on Water Supply page 2 of 58

3 UBCWM Summary comparison Conclusions...49 References...50 Appendix: Description of tables...58 CC-WaterS Climate Change and Impacts on Water Supply page 3 of 58

4 4 List of Tables Table 1: Model functionality HEC-1, HEC HMS...15 Table 2: Model complexity HEC-1, HEC HMS...16 Table 3: Model applicability HEC-1, HEC HMS...17 Table 4: Model functionality TR Table 5: Model complexity TR Table 6: Model applicability TR Table 7: Model functionality MODRAT...20 Table 8: Model complexity MODRAT...20 Table 9: Model applicability MODRAT...21 Table 10: Model functionality Rational Method...22 Table 11: Model complexity Rational Method...22 Table 12: Model applicability Rational Method...23 Table 13: Model functionality GSSHA...23 Table 14: Model complexity GSSHA...24 Table 15: Model applicability GSSHA...25 Table 16: Model functionality ZEMOKOST...26 Table 17: Model complexity ZEMOKOST...26 Table 18: Model applicability ZEMOKOST...27 Table 19: Model functionality TOPMODEL...28 Table 20: Model complexity TOPMODEL...29 Table 21: Model applicability TOPMODEL...29 Table 22: Model functionality HYDR²AC...30 Table 23: Model complexity HYDR²AC...31 Table 24: Model applicability HYDR²AC...31 Table 25: Model functionality of BROOK-Models...32 Table 26: Model complexity of BROOK-Models...33 CC-WaterS Climate Change and Impacts on Water Supply page 4 of 58

5 5 Table 27: Model applicability of BROOK-Models...34 Table 28: Model functionality of CATFLOW...34 Table 29: Model complexity of CATFLOW...35 Table 30: Model applicability of CATFLOW...36 Table 31: Model functionality of PREVAH...37 Table 32: Model complexity of PREVAH...37 Table 33: Model applicability of PREVAH...38 Table 34: Model functionality of WaSiM-ETH...38 Table 35: Model complexity of WaSiM-ETH...39 Table 36: Model applicability of WaSiM-ETH...40 Table 37: Model functionality of UBCWM...41 Table 38: Model complexity of UBCWM...41 Table 39: Model applicability of UBCWM...42 Table 40: Most important features of the hydrological models discussed in Chapter Table 41: Most important features of the hydrological models discussed in Chapter Table 42: Model functionality...58 Table 43: Model complexity...58 Table 44: Model applicability...58 CC-WaterS Climate Change and Impacts on Water Supply page 5 of 58

6 6 List of Acronyms CATFLOW CC CN DEM ERDC GSSHA HEC-1 (HMS) HEC-RAS HRU HSG IHDM MODRAT NRCS P/R-model SCS SHE TOPMODEL CATchment FLOW Climate Change Curve Number (SCS) Digital Elevation Model Engineering Research and Development Centre Gridded Surface Subsurface Hydrologic Analysis Hydrologic Engineering Centre 1 (Hydrological Modelling System) Hydrologic Engineering Centre River Analysis System Hydrological Response Unit Hydrological Soil Group Institute of Hydrology Distributed Model Modified Rational Method Natural Resources Conservation Service Precipitation/Runoff-model Soil Conservation Service Système Hydrologique Européen A TOPography based hydrological MODEL TR20 Technical Release 20 TR55 Technical Release 55 UBCWM WaSiM-ETH WMS USDA University of British Columbia Watershed Model Wasserhaushalts-Simulations-Modell - Eidgenössische Technische Hochschule Watershed Modelling System United States Department of Agriculture CC-WaterS Climate Change and Impacts on Water Supply page 6 of 58

7 7 Abstract In the frame of the Interreg IIIB-project CCWaterS, two questions have been posed by the Federal Ministry of Agriculture and Forestry, Water Management and Environment; Department IV (Silvicultural Land-use Planning, Forest Protection): a) Which models are capable of assessing climate change (CC) and forest management effects? b) To what extent can forest management effects (harvesting, tending operations, road-building, etc.) be quantified or simulated by established hydrological models? After an overview of the basics of hydrologic models, some aspects on how land-use changes can be considered in these models are discussed. Additionally some results of model comparisons in the current literature are presented. Six event-based models (HEC-1 /HEC-HMS, TR-20, Rational Method, MODRAT, GSSHA, ZEMOKOST) and eight continuous models (HYDR²AC, TOPMODEL, BROOK-models, CATFLOW, HQsim, PREVAH, WASIM-ETH and UBCWM) have been evaluated in terms of accuracy / correspondence to reality, complexity and their ability to reproduce changes in land-cover, forest management and CC. The results show, that the capability of hydrological models must always be seen in the context of their desired application. The overall best model does not exist. All of the investigated models are capable of analyzing effects of land-use changes or forest management, at least in form of scenariobased calculation. CC-WaterS Climate Change and Impacts on Water Supply page 7 of 58

8 8 1 Introduction Scope Since beginning of the year 2010 the BFW is engaged in the Interreg IIIB-project CCWaterS (Climate Change and Impacts on Water Supply) as a sub-contractor of the BMLFUW (Federal Ministry of Agriculture and Forestry, Water Management and Environment; Department IV - Silvicultural Landuse Planning, Forest Protection). CCWaterS deals with potential impacts of CC on water supply. BFW In the frame of the work package Land Uses and Water Safety knowledge on the capability of hydrological models to reproduce forest management effects, land-use change and CC-effects on catchment hydrology is of central interest. Aims The report at hand aims at discussing of following themes: 1) Analysis of existing hydrological models: Which models are capable of assessing climate change and forest management effects? 2) To what extent can forest management effects (harvesting, tending operations, road-building, etc.) be quantified or simulated by established hydrological models? Limiting basic conditions Screening of relevant models has been limited to micro-scale and meso-scale catchments. The main focus is laid on runoff development, conform to the contract for services with the awarding authority, the Federal Ministry of Agriculture and Forestry, Water Management and Environment; Department IV (Silvicultural Land-use Planning, Forest Protection). Effects of land-use and management effects on runoff and catchment discharge at the macro scale will be discussed by the IWHW; Department of Water, Atmosphere and Environment (BOKU - Vienna). This has been the result of a meeting on 8 th April 2010 at the BOKU. Specific aspects of modeling forest-water balances at the plot scale and small micro scale will be investigated by the Department of Forest and Soil Science (BOKU Vienna). Both institutions will launch separate reports. Also flooding forecast models are not discussed in this report. Such an analysis has been done by the DWA (2009) for catchments up to m² size. Structure Chapter 2 informs about some basics on hydrological models; Various aspects of considering land-use changes in hydrological models are discussed in Chapter 3; a short glimpse of model reviews in the current literature is given in Chapter 4; Evaluation of different models is presented in Chapter 5. Chapter 8 contains a short summary comparison; the final conclusions are presented in Chapter 7. An explanation of the tables presented in Chapter 4 is provided in the Appendix. CC-WaterS Climate Change and Impacts on Water Supply page 8 of 58

9 9 2 Overview of the types of hydrological models The following description of model categories follows the account in Gosain et al. (2009) and Kirnbauer et al. (2008). For more details see their reports. 2.1 Black-box models The black-box models mathematically describe the relation between input (precipitation) and output (runoff) without describing the physical process by which they are related, and establish a statistical correspondence between input and output. No black-box models are discussed in this report. 2.2 Deterministic / physical models These models are based on complex physical theory and require large amounts of data and computation time. The models are necessarily distributed because of the non-linear partial differential equations, which describe the hydrologic processes. The models offer the ability to simulate the complete runoff and the effect of catchment changes, which is particularly important in case of resource management. A noteworthy aspect of the deterministic models is that these models offer the internal view of the process which enables an improved understanding of the hydrologic system. Important representatives from this group are SHE, IHDM. 2.3 Conceptual models These models serve as a trade-off between the deterministic approach and black-box approach. Conceptual models are formulated by a number of conceptual elements, each of which is a simplified representation of one process element of the system being modeled. The conceptual models can be characterized into event models and continuous models (Gosain et al. 2009): Event models only represent a single runoff event occurring over a period of time, ranging from an hour or less to several days, depending on the size of the catchment. The initial conditions for each event must be provided as an input. Event models cannot utilise the record of soil moisture conditions of the basin in a continuous manner. Hence, event models are not useful for ungauged catchments. For event based models (like HEC-HMS, TR 20, Rational Method, MODRAT, ZEMOKOST, HYDR²AC) prevalent system conditions have to be defined before running the model. The big advantages of event-based models normally are the small number of parameters needed, easy parameterization of the models, and that these data can be collected with relatively small effort at a precise level. For instance the P/R-model ZEMOKOST uses surface runoff coefficients and surface roughness coefficients - a lot of additional information is included in these parameters (e.g. quality of substrate, vegetation effects, type and intensity of land use, management effects, effects of surface sealing) (Markart et al. 2004, 2006). Continuous models operate over an extended period of time determining flows during all periods irrespective of the magnitude of flow. Continuous models show flexibility, they are able to use bio-geographic information (like HYDR²AC, BROOK-models, HQsim, CATFLOW, WASIM- ETH, UBCWM). Their big advantage is that the definition of the antecedent conditions is not necessary, as these models react to the prevalent system conditions of the respective catchment. One downside is the large number of parameters necessary to run this type of model. Conceptual models further can be classified into lumped models and distributed models: CC-WaterS Climate Change and Impacts on Water Supply page 9 of 58

10 Lumped models In lumped models the spatial variations of watershed characteristics are generally ignored. Precipitation is considered to be spatially uniform throughout the watershed. Average values of watershed characteristics, i.e. vegetation, soils, geology or topography, are used. Spatial heterogeneities are not well reproduced by average parameters. The results produced by these models display the average watershed conditions. Such models are calibrated based on historic records; any bias existing in the data is transferred to a set of optimized parameter values. Therefore applicability of the model to other catchments is restricted. An extensive amount of data are necessary, hence lumped models are not suitable for ungauged catchments Distributed models Distributed models take the spatial variability of watershed properties into account (by Representative Elementary Area (REA), Hydrological Response Units (HRU s) or Grouped Response Units (GRU s). The distributed models are well suited for: i) Evaluating the effects of land-use change within a watershed ii) Evaluating the effects of spatially variable inputs and outputs iii) Simulating the water quality and sediment yield on a watershed basis iv) Simulating the hydrological response of ungauged catchments where no data are available for calibration. (Gosain et al. 2009) Major problems of distributed models include: large quantity of required input parameters - inefficient for everyday operational hydrology often insufficient information about catchment parameters available (question of scale) insufficient understanding of the processes of runoff generation at the catchment scale Semi-distributed models Semi-distributed models are a compromise between lumped models and fully distributed models. The model algorithms are simple, but physically based. Semi-distributed conceptual models are widespread because they offer an easy and often adequate way to simulate the discharge at the catchment outlet. Finer spatial subdivision may be necessary to describe the effects of spatially heterogeneous events like land use change or forest management measures (e.g. buffer strips or erosion control). Semi-distributed hydrological models generally have the advantages of short calculation times, comparatively low calibration needs and high model efficiency, but lack the ability to consider localization effects of land-use change (Lautenbach et al. 2006). Most of the evaluated models belong to this group. CC-WaterS Climate Change and Impacts on Water Supply page 10 of 58

11 11 3 Aspects concerning land-use changes in hydrological models At the basin scale, significant factors affecting hydrological impacts of CC include latitude, topography, geology, and land use. Under scenarios of future CC, many basins are likely to experience changes not only in their mean hydrology, but also in the frequency and magnitude of extreme hydrological events. Impacts of CC on water quality are largely determined by hydrological changes and the nature of pollutants controlled by flushing or dilution (Praskievicz and Chang 2009). Exact results of hydrological models depend, to a high degree, on correct precipitation data. Precipitation data generally lack quality even in central Europe; furthermore data quality decreases with increasing sea level, due to declining density of the gauging station network and rising effects of meteorological factors (i.e. losses due to wind drift, hail, etc. - Gattermayr 2003). The impact of land use on flood formation decreases with an increase in rainfall intensity (Wahren et al. 2007). Identifying and quantifying the hydrological consequences of land-use change are not trivial exercises, and according to DeFries and Eshleman (2004) are complicated by: i) the relatively short lengths of hydrological records ii) the relatively high natural variability of most hydrological systems iii) the difficulties in controlling land-use changes in real catchments within which changes are occurring iv) the relatively small number of controlled small-scale experimental studies that have been performed v) the challenges involved in extrapolating or generalizing results from such studies to other systems. One of the most important factors influencing the water supply of a site is canopy throughfall how much water is intercepted? Canopy throughfall depends on the leaf area index (LAI) and the number of trees per area unit. There are numerous investigations on interception of forest stands worldwide indicating that canopy interception for coniferous forest is about 4-6 mm per single rain event (Markart 2000). There are a lot of studies cited in Keim et al. (2005) indicating a limited spatial structure of throughfall in storms but hint at a specific correlation lengths under isolated trees. Throughfall data for a deciduous stand and two coniferous stands analyzed by Keim et al. showed that throughfall amounts were correlated varied by stand and season from no correlation to spatial correlation lengths of approximately 3-10 m. Changes from forests to grassland and vice-versa can be simulated by following parameter changes (Wahren et al. 2007): i) increase of root depth (a larger part of soil storage can be emptied via transpiration ii) an additional organic layer on top of the mineral soil iii) a higher amount of organic matter in the top layers of mineral soil, and iv) more macro-pores represented by a higher macro-pore conductivity. For many regions in Central Europe even the information on these four points is only available at a very rough level. To which level will the rooted area increase after afforestation? This depends on numerous factors (i.e. climate conditions, topography, altitude, quality of substrate, intensity of roads in former and current land use, type and provenience of tree species, and many others). Markart (2000) observed that the roots of a 25 years old pinus cembra afforestation did not go deeper than 35 cm, despite or even because of a good supply with nutrients and water in the upper soil. Spatially distributed modelling is an important instrument for studying the hydrological cycle, both concerning its present state as well as possible future changes in climate and land use (Viviroli et al. 2009). Feasibility of observing land-cover changes with satellite data is unprecedented and modeling capabilities for evaluating and predicting hydrological consequences of land-use change at multiple scales have advanced at a rapid rate in recent years, owing largely to technological improvements in CC-WaterS Climate Change and Impacts on Water Supply page 11 of 58

12 12 data collection and computing. These actors now focus on the hydrological impacts of feasible landuse change (DeFries and Eshleman 2004). It is currently possible to measure landscape change over large areas and determine trends in environmental condition using advanced space-based technologies accompanied by geospatial analyses of the remotely-sensed data (Kepner et al. 2006). We are able to analyze land-use changes from the past, however the problems are: a) lack of verification in ungauged basins b) fit of models assessment/quality of input parameters The quality of a hydrological simulation depends on the ability of the underlying model to describe and accurately represent the heterogeneity of such hydrological systems at the different spatial and temporal scales (Viviroli et al. 2007). Assessment of future developments in consequence of climate change must be based on climate models and coupled suitable precipitation-runoff models. This means that reliable models are needed, suitable for the conditions of a future climate and an adapted natural environment of vegetation and soil (results of Kirnbauer et al in the frame of the INTERREG IIIB-project ClimChalp). Kirnbauer et al. state that the quality of the results obtained by hydrological and runoff-models depends strongly on the type of the model and the type and quality of input variables. CC-WaterS Climate Change and Impacts on Water Supply page 12 of 58

13 13 4 Model reviews in the current literature The current literature offers a wide range of model comparisons, which differ according to the thematic focus of the publication and the choice of models. Several comparisons include the models dealt with in this publication: Beckers et al. (2009a) and the subsequent summaries of this comparison in Beckers et al. (2009b and 2009c) focus on hydrologic models for forest management and climate change applications in British Columbia and Alberta. They reviewed a total of 30 models, mainly from Northern America and Europe. The main aim was to help improve the ability of forest managers to understand the assumptions and limitations that underlie model results. The authors grouped the models by their complexity (high, medium and low) and came to the conclusion that there is no best model for operational use in forest management, but instead recommended a total of nine models from those 30. The choice between these models depends on the skill, budget and intended application of the respective modeller (Becker et al. 2009a). Holländer et al. (2009), Bermann et al. (2010) compared ten conceptually different models in predicting discharge from an artificial catchment area in North-East Germany (Chicken Creek). The main aim was to bring together hydrological modellers from across Europe and compare the modelled with the observed water balance in a three-year project. They concluded that none of the models came close to predicting the correct water balance and state, that the crucial parameters were the soil parameters and the initial soil-water content. Singh & Frevert (2002a) bring together a wide range of publications in order to provide a comprehensive overview of mathematical models of small watershed hydrology applications. The selection of the models is performed according to their representativeness, comprehensiveness, applicability, connectivity and geographical representation. V.P. Singh is the editor of a series of other publications involving the comparison of hydrologic models, including Singh (1995), Singh & Frevert (2002b) and Singh & Frevert (2006). Casper (2002) reviewed three hydrological models with respect to the most important runoff generation processes and their spatial and temporal distribution in the study area of the Duerreychbach in the Northern Black Forest, Germany. He came to the conclusion, that the employed physically-based models (CATFLOW, WaSiM-ETH and PRMS) offer a better result, than simulations based on the TOPMODEL-approach, as they are more flexible and closer to the actual processes observed in the field. Merz and Blöschl (2000) offered a market-overview and examples for P/R-modelling software. The most important features of 9 P/R-models are presented in a short overview. Use of SCS-model (USDA-SCS 1985) and models using this method for P/R-calculation in Alpine catchments can be problematic (Kohl 2010). Comparisons of the SCS-CN-approach with results from heavy rain experiments at the Federal Research and Training Centre for Forests, Natural Hazards and landscape (BFW) showed that about 1/3 of the hydrological response units can be reproduced with the SCS-method very well, 1/3 can be reproduced by an adaptation of the curve number, but the rest of the CN was not able to reflect correct runoff reaction. Also Kuntner and Burlando (2003) observed bad concordance between modelled and observed runoff in some investigated areas. According to Patt (2001) runoff is underestimated by the SCS-method in case of events with low precipitation. Also Hagen et al. (2007) found in their comprehensive comparison of different runoff formulas and simple P/R approaches that an SCS-based method did not match real runoff characteristics very well. This list is by no means comprehensive and only highlights some of the relevant publications reviewed for this model comparison. CC-WaterS Climate Change and Impacts on Water Supply page 13 of 58

14 14 5 Comparison of models The following chapter gives an overview of a wide range of hydrologic models, including those supplied with the software package WMS, as well a number of other state-of-the-art models. This selection comprises some of the most commonly used models in Austria and one example of a model developed and employed in North America (i.e. UBCWM), in order to provide an insight into a medium complexity model possibly less well-known to European modellers. This list, however, should be viewed as a compilation of a few selected models, relevant to the issue at hand, rather than a comprehensive summary of all the potentially available models. Due to the reasons stated in Chapter 2, SCS-based P/R-models are not discussed in this report, except for HEC-products (e.g. HEC HMS), because they can also be used with a runoff coefficient approach. In Switzerland two hydrological models HAKESCH (Hochwasserabschätzung in kleinen Schweizer Einzugsgebieten = Flood water assessment in small Swiss catchments) for catchments < 10 km² and - HQx_meso_CH for catchment areas between 10 and 200 km² have been developed. This package forms a multi-model approach for the assessment of flood water peak runoff (Spreafico et al. 2003). Direct application of these models in catchments outside Switzerland is not possible, because model-programs are linked to specific Swiss datasets on the one hand and on the other some approaches cannot be transferred to catchments outside of Switzerland without adjustments due to their Switzerland-specific development (Kohl 2010). There are many more models available, which are suitable for discussion of forest management effects on runoff. However many of them are at least partly based on the same approaches (e.g. CATFLOW and HYDRUS-2D after Holländer et al. (2009), as both use Richards-equation for the description of soil water dynamics and the Penman-Monteith method to determine evapotranspiration). Therefore representatives of different model types have been analyzed (lumped, semi-distributed, distributed; physical vs. empirical, few input parameters ( free models ) vs. complex models ( gun licence obligatory, after Kirnbauer pers. comm.), micro scale, meso scale, macro scale; continuous vs. event based, etc.). As a representative of the widely distributed HBV-Model (Integrated Hydrological Modelling System; Bergström 1976) - from which many clones are available - the model PREVAH (Viviroli et al. 2007) has been analyzed. Convenient hydrological models used at the macro scale, like SHE, COSERO, and others are not discussed in the frame of this publication, see Chapter WMS models The Department of Defence (DoD) Watershed Modelling System (WMS) is state-of-the-art software that integrates hydrology, hydraulics, and water quality to help civil engineers and others involved in hydrodynamic modelling to make informed decisions about watershed management. A variety of watershed models are available, each requiring different input data in a structured format. The models utilize digital terrain data to delineate watershed and sub-basin boundaries as well as computing geometric parameters used in hydrological modelling. Digital elevation models, land use, and soil data can be analyzed to determine watershed boundaries, stream networks, and hydrologic parameters necessary for model construction. WMS allows the incorporation of other data sources into the modelling process, such as imagery and radar rainfall estimates, (U.S. Army Corps of Engineers, Research and Development Center, 2008). WMS is a comprehensive graphical modelling environment for all phases of watershed hydrology and hydraulics. WMS includes powerful tools to automate modelling processes such as automated basin delineation, geometric parameter calculations, GIS overlay computations (e.g. CN, rainfall depth, roughness coefficients), cross-section CC-WaterS Climate Change and Impacts on Water Supply page 14 of 58

15 15 extraction from terrain data, etc. With the release of WMS 8, the software now supports hydrologic modelling with HEC-1 (HEC-HMS), TR-20, TR-55, Rational Method, NFF, MODRAT, OC Rational, and HSPF. Hydraulic models supported include HEC-RAS, SMPDBK, and CE QUAL W2. The program s modular design enables the user to select modules in custom combinations, allowing the choice of only those hydrologic modelling capabilities that are required (Aquaveo, 2008). In this chapter only WMS-models with relevance to runoff development, land use, forestry and climate change are discussed HEC 1 (HEC-HMS) HEC-1, developed by the Hydrologic Engineering Centre in Davis, California, has long been one of the industry-standard programs for hydrologic analysis. The HEC-1 model is designed to simulate the surface runoff response of a river basin to precipitation by representing the basin as an interconnected system of hydrologic and hydraulic components. Each component models an aspect of the precipitation-runoff process within a portion of the basin, commonly referred to as a subbasin. A component may represent a surface runoff entity, a stream channel, or a reservoir. Representation of a component requires a set of parameters which specify the particular characteristics of the component and mathematical relations which describe the physical processes. The result of the modelling process is the computation of streamflow hydrographs at desired locations in the river basin. ( The HEC-HMS model is the latest release from the Hydrologic Engineering Centre for hydrological analysis of a watershed. The new release is based on the HEC-1 model for most analytical options and routines. The WMS software allows you to export a HEC-1 model to the new HEC-HMS file format and then use HEC-HMS to run the analysis ( Model functionality HEC-1 is a single storm event, lumped parameter model, but includes several different options for modelling rainfall, losses, unit hydrographs, and stream routing. The HEC-1 interface contained within WMS makes it simple to enter and manage input data, and display analysis results ( The Hydrologic Modelling System (HEC-HMS) is designed to simulate the precipitation-runoff processes of dendritic watershed systems. It is applicable in a wide range of geographic areas for solving the widest possible range of problems. This includes large river basin water supply, flood hydrology and small urban or natural watershed runoff ( Table 1: Model functionality HEC-1, HEC HMS Approach Spatial discretisation Spatial scale (watershed size) Temporal discretisation Input data Results Physical Semi distributed Small to large basins Time steps with variable intervals (sub-daily or greater) Depending on input data Model complexity The HEC-1 model is designed to simulate the surface response of a river basin to precipitation by representing the basin as an interconnected system of hydrologic and hydraulic components. Each component models an aspect of the precipitation-runoff process within a portion of the basin, CC-WaterS Climate Change and Impacts on Water Supply page 15 of 58

16 16 commonly referred to as a sub-basin. A component may represent a surface runoff entity, a stream channel or a reservoir. Representation of a component requires a set of parameters which specify the particular characteristics of the component and mathematical relations, which describe the physical processes. The result of the modelling process is the computation of streamflow hydrographs at desired locations in the river basin (USACE, 1998). HEC HMS is a generalized modelling system capable of representing many different watersheds. A model of the watershed is constructed by separating the hydrologic cycle into manageable pieces and constructing boundaries around the watershed of interest. Any mass or energy flux in the cycle can then be represented by a mathematical model. In most cases, several model choices are available for representing each flux. Every mathematical model included in the program is suitable for different environments and under different conditions. Making the correct choice requires knowledge of the watershed, the goals of the hydrologic study, and engineering judgment ( Table 2: Model complexity HEC-1, HEC HMS Data requirements (properties of input parameters) Number 6 Type Orthophoto, DEM, land use, soil type, curve number (SCS), precipitation Accuracy / significance High (limited to few, significant parameters) Correspondence to reality Direct field measurement of parameters Effort for provision Low (due to low number of parameters) Cost-benefit of parameters High (limited to few, significant parameters) Resource requirements Processing effort GIS analysis Personnel Overall model complexity Depending on the watershed GIS analysis required Can be completed by one person Low Model applicability The main limitation of HEC-HMS appears to be the empirical manner in which evapotranspiration and snowmelt are handled. The model is not recommended for forest management applications, given that its development is emphasized on simulating river processes rather than watershed processes. The model can be applied to small to large watersheds, with gradual topography and rain or snow regimes. (Beckers et al. 2009a). Fischenich (1999) illustrated the use of HEC-HMS in the context of stream restoration. The model was used to simulate peak flows for Stirling Branch in the western US. In the contributing watershed, the average elevation was 1300 feet, while the total watershed area was 3.2 km 2. The effects of no restoration, restoration with trimmed trees in the floodplain, and restoration with trees and shrubs on peak flows were simulated. While this case study does not directly address forest management hydrology, it does examine the effects of increasing channel vegetation. (Beckers et al. 2009a) CC-WaterS Climate Change and Impacts on Water Supply page 16 of 58

17 17 Table 3: Model applicability HEC-1, HEC HMS Representation of system Physiographic changes Climatic changes Land cover / land use Forest management Simulation of processes Criteria Most parameters can be measured in the field Wide range available No specific applications to forest management Regimes simulated: nival and pluvial, wide range of hydrologic processes can be modelled (e.g. snowmelt, lakes, wetlands, groundwater flow) The model can be adjusted for alternative climate change scenarios TR-20 TR20 was originally developed by the Hydrology Branch of the US Soil Conservation Service (SCS) in cooperation with the Hydrology Laboratory, Agricultural Research Service (ARS), through a contract with C-E-I-R, Inc. numerous modifications and additions have been made since by the SCS ( It is a single event watershed scale runoff and routing model. It computes direct runoff and develops hydrographs resulting from any synthetic or natural rainstorm. Developed hydrographs are routed through stream and valley reaches as well as through reservoirs. Hydrographs from tributaries are combined with those of the main stream. Branching flow (diversion) and baseflow can also be accommodated for (USDA, NRCS: WinTR20.html). Model functionality The TR-20 software assists the modeller in hydrologic evaluation of flood events for use in analysis of water resource projects. The program is a physically-based event model which computes direct runoff resulting from any man-made or natural rainfall. It does not provide recovery of initial abstraction or infiltration during periods of no rainfall within an event ( Table 4: Model functionality TR-20 Approach Spatial discretisation Spatial scale (watershed size) Temporal discretisation Input data Results Physical Semi-distributed (refinement by DEM) Small to medium-sized basins (max. 25 km²) Non Sub-daily (hours) CC-WaterS Climate Change and Impacts on Water Supply page 17 of 58

18 18 Model complexity TR-20 The "minimum" input data set for a TR-20 model run includes storm analysis information, a reach (with cross section) and/or sub-area data, and a specification of the output. If a hydrograph is applied to the top of the reach, sub-area data is not required. ( Sub-area data includes an identifier, the drainage area, the runoff curve number (CN), the time of concentration (Tc), and the reach or the watershed outlet receiving the sub-area flow. If the sub-area CN is not available, detailed land use and hydrologic soil group (HSG) combination information can be entered for the sub-area. The TR-20 model will calculate the CN. If the time of concentration is not available, detailed flow path information can be provided (flow lengths, slopes, Manning n, and cover type of sheet and shallow concentrated flow and flow length, flow end area and wetted perimeter or flow velocity for channel flow), and the program will calculate the sub-area time of concentration ( /WinTR20.html) Reach data includes an identifier, stream cross section data (elevation, discharge, end area, friction slope, and top width), and channel length. If the valley length is different from the channel length, valley length should be provided as well. A constant base flow value can be added to the reach if appropriate, and split flow can be defined. A reach can also represent an impoundment and its associated storage in the pool area. In this case only a structure rating identifier is needed as reach data ( Table 5: Model complexity TR-20 Data requirements (properties of input parameters) Number 12 Type Accuracy / significance Correspondence to reality Effort for provision Cost-benefit of parameters Precipitation, sub-area data, reach data Few, significant parameters, partly estimated values Parameters are collected in the field and supplemented with map data Low (due to low number of parameters) High (limited to few, significant parameters) Resource requirements Processing effort GIS analysis Personnel Overall model complexity Depending on the watershed GIS analysis required Can be completed by one person Low Model applicability The program develops flood hydrographs from runoff and routes the flow through stream channels and reservoirs. Routed hydrographs are combined with those from tributaries. Procedures for hydrograph separation by branching or diversion of flow and for adding base flow are provided ( Peak discharges, their times of occurrence, water surface elevations and duration of flows can be computed at any desired cross section or structure. Complete discharge hydrographs, as well as discharge hydrograph elevations, can be obtained if requested. The program provides for the analysis of up to nine different rainstorm distributions over a watershed using various combinations of land CC-WaterS Climate Change and Impacts on Water Supply page 18 of 58

19 19 use, floodwater retention structures, diversions, and channel modifications. Such analysis can be performed on as many as 200 sub-watersheds or reaches and 99 structures in any single continuous run ( TR-20 may be used to evaluate flooding problems, alternatives for flood control (reservoirs, channel modification, and diversion), and impacts of changing land use on the hydrologic response of watersheds. (USDA, NRCS: Table 6: Model applicability TR-20 Representation of system Physiographic changes Climatic changes Land cover / land use Forest management Simulation of processes Criteria Most parameters can be measured in the field Wide range available No specific applications to forest management Regimes simulated: pluvial Changes can only be included indirectly MODRAT MODRAT is a modified rational method computer program developed by the Los Angeles County Department of Public Works (LACDPW) to compute runoff rates under a variety of conditions common to the area of Los Angeles, California. Being the successor of F0601, MODRAT contains all the features of the F0601 as well as updated capabilities for watershed modelling in the Los Angeles area. MODRAT may be used to find flow rates for any watershed with any combination of existing or proposed channels and drains. Further, the watershed may be undeveloped, partially developed, or completely developed. The model will compute runoff rates for a 50-year, 25-year, or 10-year frequency design storm (developed by LACDPW), as well as any other storm which can be represented by a rainfall mass curve. Given any combination of the above variables, MODRAT will compute a hydrograph for each subarea and mainline collection point in the watershed. ( Model functionality The basis of MODERAT is the rational formula after Mulvaney (1851), which is one of the earliest and simplest hydrological models (Gosain et al. 2009). As a method of urban hydrology, the rational method falls short in several ways. Firstly, the method does not produce a hydrograph, only a single flow rate. Secondly, the rational method does not account for changing (time dependent) conditions such as soil condition or rainfall intensity. Finally, results are not very accurate for large areas. Due to these problems, MODRAT contains the following modifications: Rainfall intensity, i, is a variable dependent on rainfall frequency, storm time, and time of concentration. The variation of i is represented by a temporal distribution curve (rainfall mass curve). C, the runoff coefficient, varies with soil type, rainfall intensity, and imperviousness. CC-WaterS Climate Change and Impacts on Water Supply page 19 of 58

20 20 The time variation of C and i allow the flow, Q, to vary with time, thus producing a hydrograph. The area under the hydrograph represents the total volume of flow from a watershed, a variable which the rational method does not provide. Hydrographs may be computed for a number of subareas, for each lateral to the main channel, and for each collection point on the main channel. These hydrographs are routed and combined as computation progresses downstream. The above modifications to the rational method allows for the computation of storm hydrographs for any size watershed. With such improvements, the modified rational method (MODRAT) has been adopted by LACDPW as the preferred method of hydrologic analysis. ( Table 7: Model functionality MODRAT Approach Spatial discretisation Spatial scale (watershed size) Temporal discretisation Input data Results Physical Semi-distributed Small to large-sized basins Sub-daily (minutes) Sub-daily Model complexity MODRAT uses a specified storm event and a time of concentration to calculate runoff at different times throughout the storm. Calculating flows based on the rainfall distribution results in a runoff hydrograph. The volume of runoff equals the area under the hydrograph curve. MODRAT allows users to route hydrographs generated in each subarea through conveyances and combine hydrographs based on time. MODRAT produces peak flows equal to or lower than the flows calculated using the rational method. The reduction in peak results from attenuation, channel storage, and combining flows that peak at different times (Hydrological Manual, 2006). Table 8: Model complexity MODRAT Data requirements (properties of input parameters) Number Less than 25 Type Accuracy / significance Correspondence to reality Effort for provision Cost-benefit of parameters E.g. Precipitation, sub-area data, reach data, soil type, Manning n low (many empirical relations) Direct field measurement of parameters, estimated data Low (due to low number of parameters) High (limited to few parameters) Resource requirements Processing effort GIS analysis Personnel Overall model complexity Depending on the watershed GIS analysis required Can be completed by one person Low complexity CC-WaterS Climate Change and Impacts on Water Supply page 20 of 58

21 21 Model applicability The model relies on many empirical relations to represent physical processes. It requires a high level of expertise for application. There is a limited public domain version and the full package is expensive. (Majumder, 2008) Table 9: Model applicability MODRAT Representation of system Physiographic changes Climatic changes Land cover / land use Forest management Simulation of processes Criteria Many estimated data Changes in land cover or land use cannot be represented in the model Calculation of scenarios Changes in forest management cannot be represented in the model Calculation of scenarios Regimes simulated: pluvial, modified rational equitation Changes can only be included indirectly calculation of scenarios Rational Method The rational method is a simple technique for estimating a design discharge from a small watershed. It was developed by Kuichling (1889) for small drainage basins in urban areas (Thompson 2006). Although it is often considered simplistic, it still is appropriate for estimating peak discharges for small drainage areas of up to about 200 acres (80 hectares) in which no significant flood storage appears ( txdotmanuals/hyd/the_rational_method.htm). The Rational Method is widely used to estimate the peak surface runoff rate for the design of a variety of drainage structures, such as the length of storm sewer, a storm water inlet, or a storm water detention basin. The Rational Method is most suitable for small urban watersheds that lack storage basins, such as ponds or swamps. It is most suitable for areas with less than 100 acres, but is sometimes also used for larger regions. (Bengston, 2010) Model functionality WMS includes an interface to the rational method which can be used for computing peak flows on small urban and rural watersheds. The interface includes the capability to combine runoff from multiple basins. Two different methods for determining peak flows/hydrographs at downstream confluences are available ( Traditionally a time of concentration is determined at a downstream confluence by determining the longest combination of time of concentration and routing travel time. Given a time of concentration for the outlet, rainfall intensity can be determined from a rainfall-intensity-duration curve and a peak flow computed. The hydrograph for the confluence is then determined in the same manner. In which they are determined for sub-basins; by using the peak flow, time of concentration, and a dimensionless hydrograph ( Alternatively, hydrographs for the sub-basins can be computed and then routed (lagged) and combined by summing up the confluence points. When using this method, retention basins may be defined at confluence points in order to determine the effect of storage on the computations. All of the computations for peak flows, hydrographs, and routing are done within WMS ( CC-WaterS Climate Change and Impacts on Water Supply page 21 of 58

22 22 Table 10: Model functionality Rational Method Approach Spatial discretisation Spatial scale (watershed size) Temporal discretisation Input data Results Physical Lumped Small basins Sub-daily (minutes) Sub-daily Model complexity The Rational Method shall be used only in cases, where the tributary area is 40 acres or less. The Rational Method is a standard method for calculating the peak runoff rate for a parcel. The results of its use are very sensitive to the coefficients selected. As a result the method is best suited for use on small parcels where the additional time that may be required to use another method may not be justified. Larger parcels should utilize more accurate methods. (Drainage Rules, B5) Table 11: Model complexity Rational Method Data requirements (properties of input parameters) Number 6 Type Precipitation, sub-area data, reach data, soil type, time of concentration, DEM Accuracy / significance Limited, due to few parameters Correspondence to reality Direct field measurement of parameters, estimated data Effort for provision Low (due to low number of parameters) Cost-benefit of parameters High (limited to few, significant parameters) Resource requirements Processing effort GIS analysis Personnel Overall model complexity Depending on the watershed GIS analysis Can be completed by one person Low complexity Model applicability Modern drainage modelling often includes detention of urban storm runoff to reduce the peak rate of runoff downstream and to provide storm water quality improvement. The Rational Method severely limits the evaluation of design alternatives available in urban and, in some instances, rural drainage design because of its inability to accommodate the presence of storage basins in the drainage area. When accommodation of any appreciable storage features in the drainage area is required, employed runoff hydrograph methods such as the NRCS Dimensionless Unit Hydrograph method ( CC-WaterS Climate Change and Impacts on Water Supply page 22 of 58

23 23 Table 12: Model applicability Rational Method Representation of system Physiographic changes Land cover / land use Forest management Many estimated data Changes in land cover or land use cannot be represented in the model; calculation of scenarios Changes in forest management cannot be represented in the model; calculation of scenarios Climatic changes Simulation of processes Criteria Regimes simulated: pluvial, rational equation Changes can only be included indirectly, calculation of scenarios GSSHA Developed with the US Army Corps of Engineers Engineering Research and Development Centre (USACE ERDC), the GSSHA model is a significant reformulation and enhancement of the CASC2D model. The GSSHA model is also fundamentally different from the CASC2D model because it extends the applicability of the model to non-hortonian basins. The CASC2D formulation assumes that once water infiltrates into the soil, it either drains vertically or is removed by evapo-transpiration. Soil water or groundwater is not considered in the context of non-hortonian runoff production. The GSSHA formulation can simulate non-hortonian runoff production ( The principal purpose of the GSSHA model is to correctly identify and realistically simulate the important hydrologic processes in watersheds. The model is intended to simulate different types of runoff production and determine the governing physical processes in watersheds, i.e. infiltration excess, saturated source areas, and groundwater discharge ( Model functionality GSSHA is a physically based, distributed-parameter hydrologic model intended to identify runoff mechanisms and simulate surface water flows in watersheds with both Hortonian and non-hortonian runoff ( It is a grid-based twodimensional hydrologic model. Features include 2D overland flow, 1D stream flow, 1D infiltration, 2D groundwater, and full coupling between the groundwater, vadoze zone, streams and overland flow. GSSHA can run in both single event and long-term modes ( Table 13: Model functionality GSSHA Approach Spatial discretisation Spatial scale (watershed size) Temporal discretisation Input data Results Physical Distributed Small to large basins Sub-daily (time step) Sub-daily (time step) Model complexity Major components of the model include spatially and temporally varying precipitation, snowfall accumulation and melt, precipitation interception, infiltration, evapotranspiration, surface runoff routing, simple lake storage and routing, unsaturated zone soil moisture accounting, saturated CC-WaterS Climate Change and Impacts on Water Supply page 23 of 58