Using Remotely Sensed Tools for Combined Drought Indicators to Enhance Drought Early Warning for Decision Support

Size: px

Start display at page:

Download "Using Remotely Sensed Tools for Combined Drought Indicators to Enhance Drought Early Warning for Decision Support"

Clinton Dickerson
6 years ago
Views:

1 Using Remotely Sensed Tools for Combined Drought Indicators to Enhance Drought Early Warning for Decision Support MARK SVOBODA, PH.D., DIRECTOR NATIONAL DROUGHT MITIGATION CENTER UNIVERSITY OF NEBRASKA-LINCOLN What on Earth Conference, Wellington, New Zealand, March 6-7, 2018

2 NDMC Program Objectives Improve the science of drought monitoring, planning, and mitigation Build awareness of drought & its impacts on society and the environment, and how human actions affect our vulnerability to drought Focus the attention of policy makers on the importance of drought policy and planning in the wise stewardship of natural resources Develop, conduct and maintain operational tools, research, outreach and training End-to-End : Research-Applications-Outreach-Operations Continuum

NDMC International Activities UN organizations: FAO, ISDR, UNDP and CCD World Meteorological Organization (WMO) USAID, World Bank Global Water Partnership

3 NDMC International Activities UN organizations: FAO, ISDR, UNDP and CCD World Meteorological Organization (WMO) USAID, World Bank Global Water Partnership (Integrated Drought Management Program) Various regional and national drought centers Numerous government agencies and universities in different countries

4 Some Context Build a combined drought indicator (CDI) system that can identify potential drought hot spots particularly in countries having limited in situ data networks Augment in situ networks with remotely sensed inputs Build drought early warning and information systems (DEWIS) through open source architecture aimed at developing local capacity and transfer of ownership to the countries/locals Inform decision makers in the areas of drought stressors on food, water and energy security.thus national security!

5 Types of Drought Meteorological Environmental/Ecological Agricultural Hydrological Socioeconomic There are indices and indicators used to identify all of these types of drought at various thresholds There is no single definition of drought Thus, in most cases, there is no one-size-fits-all drought indicator or index

6 Approaches to Drought Assessment Single index or indicator (parameter) Multiple indices or indicators Assessed stand-alone Composite/Combined (or hybrid ) Indicator Blended approach

7 Emerging Satellite-based Observations and Products Over the past 10+ years, a number of satellite remote sensing-based tools and products characterizing different parts of the hydrologic cycle that influence drought conditions allowing new composite drought indicators to be developed. Examples Evaporative Stress Index (ESI) Quick Drought Response Index (QuickDRI) Evaporative Demand Drought Index (EDDI) GRACE soil moisture and groundwater anomalies Vegetation Drought Response Index (VegDRI)

8 The U.S. Drought Monitor (USDM) and Objective Blend Drought Indicators (OBDI) were early (perhaps the earliest?) examples of multiindicator/cdi approaches in the U.S./world (Svoboda et al. 2002, BAMS)

9 Can explore and tailor multiple CDIs for different types of droughts as well as regional and seasonal differences

10 International CDI Activities MENA New Zealand Germany Czech Republic Greater Horn of Africa BraziI JRC-EDO IWMI-South Asia

12 Tunisian Composite Drought Index (TCDI)

13 Prototype Tunisian Combined Drought Index (TCDI) initial satellite-derived input components (all at 5km resolution) and weightings**: Standardized Precipitation Index 2-mo (CHIRPS2): 40% Evaporative Stress Index (ESI): 20% Soil Moisture (Land Information System): 20% Normalized Difference Vegetation Index (NDVI-anom): 20% **These are only initial, non-validated expert judgement inputs and weightings that need to be validated to determine how they match the conditions, impacts and reality of the situation the ground at the local level.feedback and other statistical + data mining techniques will help fine tune the TCDI

14 CDI/Component Validation Approaches Multi-pronged approach is best: 1. Evaluate past droughts (and wet periods) to see how the input parameters and CDI performed historically Perception vs. reality.memory tested on older droughts Work with users to test against impact data if available Social and environmental 2. Evaluation of current conditions in real-time Fresh impacts/memories 3. Machine Learning: Data mining or PCA approaches to determine the best combinations of weightings

15 Freely available on-line: handbook-drought-indicators-andindices/

16 Meteorological Based Page Ease of Use Inputs Required Additional Information Aridity Anomaly Index (AAI) Green P,T, PET, ET Operationally available for India Deciles Green P Ease of calculation and examples from Australia useful Keetch-Byram Drought Index (KBDI) Green P,T KBDI calculations are based upon the climate of the area of interest Percent of Normal Precipitation Green P Simple calculations Standardized Precipitation Index (SPI) Green P Weighted Anomaly Standardized Precipitation (WASP) Green P,T The WMO highlighted the SPI as a starting point for meteorological drought monitoring Uses gridded data in monitoring drought in tropical regions Aridity Index (AI) Yellow P, T Can also be used in climate classifications China Z Index (CZI) Yellow P Intended to improve upon what the SPI provides Crop Moisture index (CMI) Yellow P,T Weekly values are needed Drought Area Index (DAI) Yellow P Gives an indication of how the monsoon season perform Drought Monthly temperature and precipitation Reconnaissance Index needed (DRI) Yellow P, T Effective Drought Index (EDI) Yellow P Hydro-Thermal Coefficient (HTC) Yellow T,P NOAA Drought Index (NDI) Yellow P Palmer Drought Severity Index (PDSI) Yellow P,T, AWC Palmer Z Index Yellow P,T, AWC Program is available through direct contact with originator Ease in calculations and several examples in Russia Best used in agricultural applications Not green due to complexity of calculations and the need for serially complete data One of the many outputs of the Palmer Drought Severity Index calculations Rainfall Anomaly Index (RAI) Yellow P Serially complete data required Self-Calibrated Palmer Drought Severity Index (sc- PDSI) Yellow P,T, AWC Not green due to complexity of calculations and serially complete data needed Standardized Anomaly Index (SAI) Yellow P Point data used to describe regional conditions Standardized Precipitation Evapotranspiration Index (SPEI) Yellow P, T Agricultural Reference Index for Drought (ARID) Red P,T, Mod Crop Specific Drought Index (CSDI) Red P,T,Td,W,Rad,AWC,Mod,cr op data Reclamation Drought Index (RDI) Red P,T,S,R, SF Soil Moisture Based Ease of Use Inputs Needed Soil Moisture Anomaly (SMA) Yellow P,T, AWC Evapotranspiration Deficit Index (ETDI) Red Mod Soil Moisture Deficit Index (SMDI) Red Mod Soil Water Storage (SWS) Hydrological Based Red Ease of Use AWC,RD,ST,S WD Inputs Needed Serially complete data required, output similar to the SPI but with a temperature component Regionally produced in the southeastern United States and not tested widely outside of the region Quality data of many variables needed, making its use a challenge Similar to the SWSI, but contains a temperature component Additional Information Intended to improve upon the water balance of the PDSI Complex calculations with multiple inputs needed Weekly calculations at different soil depths, complicated to calculate Due to variation in both soil and crop types, interpolation over large areas is challenging Additional Information

17 Final Thoughts: CDI: Convergence of Evidence approach allows for: No one-size-fits-all, thus applying an ensemble-like approach Different thresholds from different combinations of inputs is often best Don t want to Cry Wolf early or all clear, too soon! Decision makers want ONE map, not multiple maps Annual User Forums and stakeholder engagements tell us this repeatedly However, scientists like MANY maps! Multiple CDI (regional/seasonal/sectoral-thematic) can be tested or made operational depending on the need and ability to validate them Machine Learning (e.g. PCA, Data Mining) to explore CDI input parameter relationships/weighting Being automated doesn t make it objective or better w/out validation!

18 Take away message The linkages between drought, water and food security, energy and climate change illustrates a nexus of complex problems.and solutions are going to depend on a combination of technology and integrated information that only partnerships and networks (e.g. DEWIS) can provide.

19 Thank You! Questions? Contact: Mark Svoboda

Mixing Up a Drought Indicator Cocktail, Blended, not Stirred: A Combined Drought Indicator Approach

Mixing Up a Drought Indicator Cocktail, Blended, not Stirred: A Combined Drought Indicator Approach Mark Svoboda, Ph.D., Director and Brian Fuchs, Monitoring Program Area Leader National Drought Mitigation