South Asia Drought Monitoring System (SADMS) A Collaborative project by IWMI, GWP and WMO under Integrated Drought Management Programme

Size: px
Start display at page:

Download "South Asia Drought Monitoring System (SADMS) A Collaborative project by IWMI, GWP and WMO under Integrated Drought Management Programme"

Transcription

1 South Asia Drought Monitoring System (SADMS) A Collaborative project by IWMI, GWP and WMO under Integrated Drought Management Programme NDVI 2002 NDVI 2003 Giriraj Amarnath, Niranga Alahacoon, Peejush Pani Gurminder Bharani, Vladimir Smakhtin International Water Management Institute (IWMI), Sri Lanka C. Jeganathan & Saptarshi Mondal, Birla Institute of Technology (BIT), India Singh T.P. Symbiosis Institute of Geoinformatics, India

2 Where we are based: The Hague Tashkent Washington, DC Lahore Ouagadougou Cairo New Delhi Anand Hyderabad Addis Ababa Kathmandu Vientiane Myanmar Colombo Pretoria

3 Presentation Outline Facts and Figures on major droughts in South Asia; Project objectives and Research Approach; Experimental South Asia Drought Monitor with specific examples; Operational drought monitoring System and 2015 evaluation in Sri Lanka and India; DMS Portal and its features; Next steps and way forward

4 Drought vs. Food Insecurity No commonly accepted definition of drought; No physically measurable drought variable Complex, multi-causal natural disaster with large spatial/temporal extent Efficient food security monitoring requires integration of environmental information and socio-economic information; Not existing yet Key variables (e.g. rainfall, soil moisture) can be detected via space-based sensors (data availability on a global scale, free of charge)

5 Facts and Figures Drought is one of the major constraints affecting the food security and livelihoods of more than two billion people; In SA, drought occurs frequently in arid and semiarid regions. From early 2000 onwards, severe droughts affected vast areas of SA, including western India, and southern and central Pakistan. SA regions have been among the perennially drought-prone regions of the world. India, Pakistan and Sri Lanka have reported droughts at least once in every three years in the past five decades, while Bangladesh and Nepal also suffer from frequent droughts. Key facts on Asia s drought disasters between 1980 and Number of drought events: 120 Number of people killed: 5,308 Population affected: 1.3 billion Total damages: USD billion Major drought events: 1987, 1999, 2000, 2002, 2009, 2014

6 Drought Monitoring Objective Develop an Integrated Drought Severity Index that links rainfall, temperature and vegetation health and soil moisture; Evaluate the added-value of satellite-derived datasets in data-scarce regions; Capacity building, customization for national needs and dissemination of the monitoring product; Policy and planning for drought preparedness and mitigation across sectors; and Develop framework to address Drought risk reduction and response; Develop and test a mobile phone application to collect, share and analyze socioeconomic information* Analyze the performance of seasonal forecasts*

7 IWMI s Drought Monitoring Framework Current Drought Risk Satellite-derived rainfall, temperature, soil moisture Future Drought Risk Ensemble predictions of rainfall, soil moisture, runoff Socioeconomic vulnerability Collection of socioeconomic information Calculate Monitoring Integrated Drought Severity Index (IDSI) Calculated Forecasted Drought Severity Visualization and analysis of monitoring/forecasted drought condition and socio-economic conditions Smartphone Application Satellite Vegetation Condition Satellite-derived rainfall, temperature, soil moisture Access to Monitoring and Forecasting Drought Condition

8 Drought Monitoring Approach INPUT DATA MOD09A1 (500m Surface Reflectance) MOD11A2 (1km Land Surface Temperature) TRMM 3B42 (27km precipitation estimates) GLOBELAND (30m) LULC 2014 (56m) Waterbody Mask (30m) Red Band NIR Band LST Band extraction 8 day accumulation Lag calculation (previous + current) Multi-source Merging DATA PRE- PROCESSING NDVI Calculation Time Series creation NDVI LST Composite of long-term stack images NDVI LST TRMM Time-Series Precipitation Scale and projection matching Noise reduction & Fourier smoothing NDVI LST Identification of Long-term anomalies NDVI LST TRMM Mask of Agriculture and non-agriculture areas DROUGHT MONITORING Standardized drought index calculation VCI TCI PCI DROUGHT ASSESSMENT Ground truth Validation Historical database and maps Development of Integrated Drought Severity Index (IDSI) Map and statistics generation Notes: MOD09A1 MODIS Surface Reflectance of every 8-Day product at 500m resolution; MOD11A1 MODIS Land Surface Temperature (LST) daily product at 1,000m resolution; TRMM Tropical Rainfall Measuring Mission; LULC NRSC Land Use and Land Cover from National Remote Sensing Centre; Water body mask from Landsat images; NDVI Normalized Difference Vegetation Index; VCI Vegetation Condition Index; TCI Temperature Condition Index (TCI); Precipitation Condition Index (PCI), IDSI Integrated Drought Severity Index

9 NDVI NDVI variability for countries in South Asia Year Sri Lanka India Pakistan Afghanistan

10 Calculation of Drought Monitoring Indices Index: Vegetation Condition Index (VCI) Data : MODIS Surface Reflectance Spatial: 500m Temporal: Every 8-day Where, VCI n = NDVI n NDVI LT_min NDVI LT_max NDVI LT_min VCI n = Vegetation Condition Index of an 8 days composite NDVI n = Mean Normalized Difference Vegetation Index off current and previous composite n = 8 days composite NDVI LT_max & NDVI LT_min = Long term ( ) max & min of NDVI n VCI is an indicator on the status of the vegetation cover as a function of the NDVI minimum and maximum. Also, VCI values indicate how much the vegetation has progressive or declined in response to weather. It was concluded that VCI has provided an assessment of spatial characteristics of drought. The 8-day NDVI is been layer stacked and used in the study. April, May, June, July, August and September of every year from 2001 to 2014 is been grouped in mean, then each pixel s minimum and maximum can be used to derive the vegetation conditional index

11 Vegetation Condition Index (VCI) for Sri Lanka Weekly composite June 2, 2013 June 4, 2013 July 2, 2013 July 4, 2013 August 3, 2013 November 1, 2013 October 3, 2013 October 1, 2013 September 3, 2013 September 1, 2013

12 Calculation of Drought Monitoring Indices Index: Temperature Condition Index (TCI) Data : MODIS Land Surface Temperature Spatial: 1000m Temporal: Every 8-day TCI n = LST max LST LST max LST min Where, T is brightness temperature. Maximum and minimum T values are calculated from the long-term record of remote sensing images for a period of Low TCI values indicate very hot weather. TCI, a remote sensing based thermal stress indicator is proposed to determine temperature-related drought phenomenon TCI assumes that drought event will decrease soil moisture and cause land surface thermal stress; TCI algorithm is similar to the VCI one and its conditions were estimated relative to the maximum/minimum temperature in a given time series. However, opposite to the NDVI, high LST in the vegetation growing season indicates unfavorable conditions while low LST indicates mostly favorable condition

13 Temperature Condition Index (TCI) for Sri Lanka 2001 Weekly composite 121 May1st week 201 July 3 rd week 225 Aug 2 nd week 241 Aug-4 th Week 305 Nov 1 st week 281 Oct 3 rd week 273 Oct 1 st week 257 Sep 2 nd week

14 Calculation of Drought Monitoring Indices Index: Precipitation Condition Index (PCI) Data : TRMM 3B42 Spatial: 0.25degree Temporal: Daily (accumulated rainfall rate) TCI n = TRMM TRMM min TRMM max TRMM min Where, TRMM, TRMM max and TRMM min are the pixel values of precipitation and maximum, minimum of it respectively in daily during TRMM data provides meteorological drought information and has spatial and temporal climate component but it cannot be directly analyzed with VCI and TCI. PCI was normalized by the TRMM 3B42 data using a similar algorithm of VIC to detect the precipitation deficits from climate signal. PCI also changes from 0 to 1, corresponding to changes in precipitation from extremely unfavorable to optimal. In case of meteorological drought which has an extremely low precipitation, the PCI is close or equal to 0, and at flooding condition, the PIC is close to 1.

15 Precipitation Condition Index (PCI) for Sri Lanka 2001 Weekly composite 121 May1st week 201 July 3 rd week 225 Aug 2 nd week 241 Aug-4 th Week 305 Nov 1 st week 281 Oct 3 rd week 273 Oct 1 st week 257 Sep 2 nd week

16 Integrated Drought Response Index (IDSI) Sri Lanka VCI TCI PCI 1 st Week September 2013 IDSI VCI Vegetation Condition Index; TCI Temperature condition Index; PCI Precipitation condition Index; IDSI Integrated Drought Severity index IDSI ijk = L VCI ijk c + 1 (L (VCI ijk +TCI ijk +PCI ijk +c) TCI ijk + PCI ijk Where, IDSI ijk, VCI ijk, TCI ijk and PCI ijk are IDSI, VCI, TCI and PCI value for pixel i in composite j of year k. The IDSI value ranges between 0 to 100; L is the normalization factor to keep the output value in expected range and c is a constant to avoid null in denominator. The values close to 0 reveals extreme drought situation as vegetation is under stress, precipitation is very low and temperature is very high. Likewise, the values closer to 100 reveals normal situation as vegetation growth is good, precipitation is high and temperature is favourable.

17 VCI, TCI & IDSI VCI, TCI & IDSI Intercomparison of drought indicators (Precipitation, Temperature, Vegetation and IDSI) Rainfall Rainfall (mm) (mm) Gujarat, India 100 Inter-relationship Between Different Indices (VCI, TCI, RF & IDSI) Rainfall VCI TCI IDSI

18 Characterizing Drought Severity Category Description Possible impacts IDSI Ranges D4 DS Extreme Exceptional and widespread crop/pasture losses Shortages of water in reservoirs, streams, and wells creating water emergencies < 5 (with very low values of VCI, PCI and TCI) D3 DS Moderate Major crop/pasture losses Widespread water shortages or restrictions 5 10 (with low values of VCI, PCI and TCI) D2 DS Severe Crop or pasture losses likely Water shortages common Water restrictions imposed D1 Stress Some damage to crops, pastures Streams, reservoirs, or wells low, some water shortages developing or imminent Voluntary water-use restrictions requested D0 Watch Going into drought: short-term dryness slowing planting, growth of crops or pastures Coming out of drought: some lingering water deficits pastures or crops not fully recovered (with moderate values of VCI, low PCI and TCI) (with moderate VCI, low PCI and moderate TCI) (with moderate values of VCI, PCI and TCI) Normal Normal >40 (vegetation growth is normal with essential variables with a function of high VCI-TCI-PCI) Healthy Healthy >60 (vigor vegetation with strong correlation on climate indicators)

19 SL Disaster Management Centre (DMC) Drought Maps

20 Spatio-temporal (yearly) pattern of drought severity index for drought years (2012 and 2014) and good year (2015) Area (sq. km) week 16 week 17 week 18 week 19 week 20 week 21 week 22 week 23 week 24 week 25 week 26 week 27 week 28 week 29 week 30 week 31 week 32 week 33 week 34 week 35 week 36 week 37 week May June July August September October Drought year (2012 and 2014) Good year (2015)

21 VCI, PCI and Drought Indices for drought year (2009) and normal year (2010), Rajasthan Long term NDVI records for 15years at each 8-day products of rain-fed season (July September) was used in the present study. Two consecutive years, 2009 and 2010 were chosen for their distinct NDVI characteristics. Good correlation observed between the precipitation, vegetation condition and IDSI for drought year and normal year.

22 VCI, PCI and Drought Indices for drought year (2001) and normal year (2007), Andhra Pradesh High correlation observed between 3- month SPI, IDSI and rice crop production

23 VCI, PCI and Drought Indices for drought year (2001) and normal year (2007), Maharashtra

24 VCI, PCI and Drought Indices for drought year (2009) and normal year (2010), Rajasthan Rajasthan Maize VCI vs. Crop Yield Crop type: Maize SPI vs. IDSI YAI vs. IDSI Average VCI of rain-fed season was compared with yield of major rain-fed (kharif) crops which reveals that a good agreement 3-month SPI also had a good correlation with IDSI for drought year and normal year High correlation co-efficient (r) was found to be 0.71, 0.72 and 0.71 (p = 0.05) for sorghum, pearl millet and maize respectively which reveals that there is a strong positive correlation present between VCI and yield of major kharif crops

25 VCI, PCI and Drought Indices for drought year (2001) and normal year (2007), Sindh Province, Pakistan Long term NDVI records for 15years at each 8-day products of rain-fed season (July September) was used in the present study. Normal year (2007) and drought year (2001) characterize the effects of precipitation on vegetation condition and drought severity. Drought affected districts : Tharparkar, Mirpur Khas, Sanghar, Tando Allahyar

26 VCI, PCI and Drought Indices for drought year (2014) and normal year (2014), Sri Lanka Area sq.km Spatio-temporal variation of drought extent for Sri Lanka ( ) Major drought events are 2009, 2011, 2012 and 2014 Districts: Moneragala, Kurunegala, Hambantota, Moneragala, Anuradhapura and Ampara

27 week 16 week 18 week 20 week 22 week 24 week 26 week 28 week 30 week 32 week 34 week 36 week38 week 16 week 18 week 20 week 22 week 24 week 26 week 28 week 30 week 32 week 34 week 36 week38 week 16 week 18 week 20 week 22 week 24 week 26 week 28 week 30 week 32 week 34 week 36 week38 week 16 week 18 week 20 week 22 week 24 week 26 week 28 week 30 week 32 week 34 week 36 week38 Polonnaruwa 100% 80% 60% 40% 20% 0% IDSI Product ( ) Sri Lanka % 80% 60% 40% 20% 0% 2012 May June July August SeptemberOctober May June July August SeptemberOctober 100% % % 80% 60% 60% 40% 40% 20% 20% 0% 0% May June July August SeptemberOctober May June July August SeptemberOctober Highlights scale of drought severity from drought is much lower compare to 2012 and 2014 (note exception rainfall in NE province during the rainy season (Dec 2010 Feb 2011)

28 2015 Bundelkhand Drought, India

29 2014 DROUGHT IN MAHARASHTRA STATE, INDIA Groundnut and other crops impacted due to lack of water supply in OS District Fields of Sorghum affected by drought in Solapur District Drinking water supply in drought affected areas of Beed district Grape fields affected by drought in Osmanabad District

30 2015 DROUGHT IN MAHARASHTRA STATE, INDIA

31 2015 DROUGHT IN MAHARASHTRA STATE, INDIA

32 Key remarks An operational platform that integrates various drought products to provide advanced drought monitoring and assessment information for various purposes A first regional platform for South Asia and have inherently finer spatial detail (500m resolution) than other commonly available global drought products

33 Possible collaborative initiatives with Regional Partners in South Asia Operational drought monitoring system setting up in target countries the basis for coordination for drought mitigation effort. Historical and current high spatial and temporal resolution drought risk and propagation mapping online; Identified hot-spot areas where droughts are more intense and frequent; State/National capacity in drought monitoring built in all participating agencies to address the gaps in Drought Risk Reduction; Explore how IDSI product can be used to explore in insuring the farmers rather than just relying only on rainfall data; Cooperation with agencies in setting up socio-economic drought vulnerability integrated with IDSI product; Develop and test a mobile phone application to collect, share and analyze socio-economic information

34 THANK YOU