1. AQUEDUCT Charles Iceland Use of Geo and Satellite Data
September 5, 2013

2. WATER
STRESS

3. Baseline Water Stress 2010 BWS = 2010 total withdrawals / mean(Ba)
Ensemble_view.R
BWS = 2010 total withdrawals / mean(Ba)
mean(Ba) calculated using mean annual NASA GLDAS-2/NOAH runoff from

4. Aqueduct water supply estimates
NASA Global Land Data Assimilation System (GLDAS) plays a key role:
GLDAS inputs include:
Temperature
Precipitation
Elevation
Wind speed
Water retention of soil
Etc.
GLDAS outputs include:
Soil moisture
Evapotranspiration
Runoff (surface and shallow groundwater)
GLDAS runoff values for period are used to bias-correct runoff estimates from 6 GCMs

5. Bias-correcting model runoff
Make Burlington plots on same Y-axis?
From pixel_ts_plot.R, pixel_gev_plot.R

6. Change in total water supply
2040 relative to 1995 baseline
DRAFT

7. Total Blue Water (Bt) 21-year window

8. Change in inter-annual variability of water supply
2040 relative to 1995 baseline
DRAFT

9. Interannual Variability (IAV) 21-year window

10. Change in seasonal variability of water supply
2040 relative to 1995 baseline
DRAFT

11. Seasonal Variability (SV)

12. Projected Water Stress 2020
DRAFT
Ensemble_view.R
Water stress = 2020 projected total withdrawals / Ba
Ba calculated using median of 6 mean annual GCM runoff from

13. Change in water stress for 2020
relative to 2010 baseline
DRAFT

14. GROUND-
WATER

15. Groundwater Stress
the ratio of groundwater withdrawal relative to the recharge rate to aquifer size; values above one indicate where unsustainable consumption could affect groundwater availability and dependent ecosystems
Groundwater stress is groundwater withdrawal / relative to the groundwater recharge rate. The withdrawal data is from The recharge rate is modeled from data between There are 748 aquifers. Values above one indicate where unsustainable water consumption could affect groundwater availability and dependent ecosystems.
Many places around the world depend on withdrawals from underground aquifers to meet their water needs. Many of these aquifers filled gradually over the course of millennia. The red parts of this map are places where groundwater is being pumped out from aquifers faster than they are being recharged by surface water gradually seeping in. In these places, as aquifers are depleted faster than they are refilled, wells must be dug deeper and deeper, and could eventually run dry.
Data Sources:
Water Balance of Global Aquifers Revealed by Groundwater Footprint, Gleeson, T., Wada, Y., Bierkens, M.F.P., and van Beek, L.P.H.,

16. GROUNDWATER DATA
Gravity Recovery and Climate Experiment (GRACE)

17. SURFACE
WATER

18. The Global Reservoir and Lake Monitor (GRLM)
Charon Birkett, ESSIC/UMD
Curt Reynolds, USDA/FAS
A NASA/USDA sponsored program in collaboration with NASA/GSFC and the University of Maryland at College Park.
Additional lake databases and
web links.
LAKENET
Additional
3-D
imagery provided by USGS
Application of Satellite Radar Altimetry for surface water level monitoring.
Jason-2/OSTM
C.Birkett ESSIC/UMD

19. FLOODS

20. FLOOD IS GROWING BY 2050: +2.0 BILLION vulnerable to flooding
A COSTLY RISK
IS GROWING
BY 2050:
+2.0 BILLION vulnerable to flooding
+$ BILLION/YR adaptation cost
Source: Munich Re, Topics Geo. Natural catastrophes 2012

21. PREDICTIVE POWER LET’S BUILD 1KM FLOOD MAPS RIVER FLOOD MODELS
LOSS ESTIMATES
SCENARIO ANALYSIS
PROBABILITY
OF LOSS

22. DROUGHT

23. Near real-time Global Agricultural Monitoring System (GLAM)
Correlates significant anomalies to drought conditions and shortfalls in crop production.
Famine Early Warning System Network (FEWS NET)
Provides early warning on emerging and evolving food security issues.
GLAM is a collaboration between NASA/GSFC, USDA/FAS, SSAI, and UMD Department of Geography
FEWS NET is funded by USAID – partners include NOAA, USGS, NASA, Chemonics, and USDA/FAS

24. Long-term projections for drought
Projections of changes in the frequency, duration and severity of drought relative to recent experience 
Projections will be developed for multiple types of drought:
Soil moisture
Evapotranspiration deficit
Hydrological drought
Image: IPCC Fourth Assessment Report: Climate Change 2007

25. WATER
QUALITY

26. WATER QUALITY CHLOROPHYL PHOSPHORUS TURBIDITY MODIS
250m+ / twice per day
1999-
LANDSAT
30m+ / 16 days + tasked
1972-

27. Charles Iceland
Senior associate

28. APPENDIX
SLIDES

29. Aqueduct water supply estimates
NASA Global Land Data Assimilation System (GLDAS) plays a key role:
GLDAS inputs include:
Temperature
Precipitation
Elevation
Wind speed
Water retention of soil
Etc.
GLDAS outputs include:
Soil moisture
Evapotranspiration
Runoff (surface and shallow groundwater)
GLDAS runoff values for period are used to bias-correct runoff estimates from 6 GCMs
Baseline
Supply = median of mean annual runoff from 6 bias-corrected GCMs for a window of time ending in 2010
Future
Supply = median of mean annual runoff from 6 bias-corrected GCMs for a window of time centered on 2020

30. Bias-correcting model runoff
“quantile mapping” aka “cumulative distribution function matching” (Mason, 2007)
Bias correction occurs at the pixel level for each month
Based on generalized extreme value distribution (3 parameters)
Corrects for all moments, including location, spread, skew
Assumes stationarity of bias
From pixel_ts_plot.R, pixel_gev_plot.R

31. Bias-correcting model runoff
Make Burlington plots on same Y-axis?
From pixel_ts_plot.R, pixel_gev_plot.R

32. Example locations bias-corrected raw runoff
Ensemble median
GLDAS-2
Runoff (m)
Possibly adjust the Y-axes (maybe not the same, but possibly all starting at zero)?
From pixel_ts_plot.R
Year
11 yr running means

33. GOALS & MILESTONES
Objective: Project change (from baseline) in water risk for three Aqueduct Framework indicators
Water stress (Water withdrawal ratio)
Inter-annual variability
Seasonal (i.e., intra-annual or monthly) variability
Interim results: May 2013
Preliminary projections for 2020
One draft scenario of supply and demand
Six climate models; one initial condition per model
Final release: January 2014
Three time periods centered on 2020, 2030, and 2040
Three scenarios of supply and demand
Six climate models; multiple initial conditions per model

34. Baseline Water Stress Definition:
Total Annual Withdrawals / mean(Annual Available Blue Water)
Available Blue Water = accumulated runoff accumulated consumptive use
Interpretation:
The degree to which freshwater availability is an ongoing concern.
High levels of baseline water stress are associated with:
Increased socioeconomic competition for freshwater supplies,
More reliance on engineered water supply infrastructure,
Heightened political attention to issues of water scarcity, and
Higher risk of supply disruptions.

35. Change in Water Stress Definition:
Future Water Stress / Baseline Water Stress
Interpretation:
Estimated rate of change in water stress due to:
Changes in use due to population growth, economic development, and technology
Changes in supply due to climate change
High rates of change associated with:
Faster pace of socio-economic and technological change required to keep pace

36. Choosing Global Climate Models (GCMs)
Select subset of 6 models from the Coupled Model Intercomparison Project Phase 5 (CMIP5; to be used for IPCC AR5)
Selection criteria:
Availability: terms of use, parameter availability (runoff and evapotranspiration)
Quality for this purpose: best representations of historical runoff (not global mean temperature)
Long-term average
Standard deviation
Data provided by Alkama et al. (2013); evaluated 15 CMIP5 models against gauge data for 18 large basins.

37. Choices Rank Model Center Country Resolution Historical members
RCP4.5 members
RCP8.5 members
Commercial use? 1
GFDL-ESM2M
GFDL
USA
2.5 x 2  1 Y 2
MPI-ESM-LR
MPI
Germany
1.875 x 1.875 3
MRI-CGCM3
MRI
Japan
1.125 x 2.25 5 N 4
CNRM-CM5
CNRM
France
1.4 x 1.4 10
CCSM4
NCAR
1.25 x 6
INM-CM4
INM
Russia
2 x 1.5 7
CanESM2
CCCMA
Canada
2.8 x 2.8 8
IPSL-CM5A-LR
IPSL
3.75 x 1.875 9
HadCM3
MOHC UK
3.75 x 2.5
 10
none
MIROC5
MIROC  5 11
NorESM1-M
NCC
Norway
2.5 x 1.875 12
CSIRO-Mk3.6.0
CSIRO
Australia 13
FGOALS-g2.0
 LASG
China 5  14
GISS-E2-R
GISS 15
BCC-CSM1.1
BCC

38. Example locations flow accumulated runoff (Bt)
Ensemble median
GLDAS-2
Runoff (m)
Adjust Y-axes?
From pixel_ts_plot.R
Year
11 yr running means

39. Estimating water use: previous work (Coca-Cola)
Domestic Use
Industrial Use
Agricultural Use
$15,000
$60,000
$1,000
Domestic = f(population, GDP/capita) Adjusted R2=0.85
Industrial = f(GDP, GDP/Capita) Adjusted R2=0.70
Agricultural = f(population, GDP/Capita, ag land, %ag land under irrigation) Adjusted R2=0.90
Each sector responds differently to changing levels of economic development (GDP/Capita)
Cross-sectional analysis generally produces optimistic Kuznets curves

40. Preliminary maps of projected change
Baseline
Supply = mean annual runoff from GLDAS-2/NOAH current release
Demand = 2010 use
FAO Aquastat withdrawals by sector, estimated for 2010 using a mean of fixed and random effects models
consumptive use computed by consumptive use ratio (Shiklomanov and Rodda 2003)
Future
Supply = median of mean annual runoff from 6 GCMs
Demand = projected change in 2010 use
change in scenario use by sector applied to baseline use
[2010 use] * [2020 scenario use] / [2010 scenario use]
Projected change maps are computed as future / baseline