1. Augmenting Hydro-MET Data Demands of Impact Assessment Models Team: IWMI (Charlotte, Solomon) Cornell (Tamo, Dan, Zach) BDU (Seifu, Esayas)

2. Background Modeling landscape processes requires detailed climatic and geographic datasets Meteorological stations in most parts of Africa are very sparse and most watersheds are un-gauged Climate records are incomplete; high percentage of missing data and relevant variables Poor data accessibility due to lack of data sharing agreement among trans-boundary riparian countries  High-resolution global reanalysis data for SWAT modeling applications in Africa

3. Global Climate Data Sources ProductSpatial Resolution Temporal Resolution Time Horizon Variables CMAP2.5 degreeMonthly1979 – 2009Precipitation GPCP2.5 degreeMonthly1979 – 2010Precipitation GPCC0.5 degreeMonthly1951 – 2010Precipitation WorldClim30 secMonthlyClimatologyPrecipitation Min, mean and max temp TRMM 0.25 degree3-hr, Daily1997 – 2012Precipitation R1, R22.5 degree6-hr, Daily1948 – 2012Climate variables ERA402.5 degree6-hr, Daily1979 – 2001Climate variables CFSR0.312 degree (~38km) Hourly, Daily1979 – 2012Climate variables, and Hydrological quantities

4. Climate Forecast and Reanalysis System (CFSR) A coupled atmosphere-ocean-land-sea ice system Assimilates satellite radiances instead of the retrieved temperature and humidity values Accounts for changing CO2 and other trace gases, and solar variations Finer spatial (~38km) and temporal (hourly and daily) resolution Assimilates hydrological quantities from a parallel LSM forced by CMAP and CPC unified daily gauge analysis

5. Weather generator files for areas with missing and incomplete climate datasets –Solar radiation, relative humidity, wind speed –Maximum half-hour rainfall SWAT weather input files for un-gauged watersheds Climate downscaling and bias correction Study of large-scale water and energy fluxes  Applications of CFSR Data

6. CFSR Data for SWAT Modeling CFSR VariableCFSR Unit Conversion FactorSWAT Unit Precipitation RateKgm -2 s -1 x 86,400mm/day Min Temperature at 2mK- 273.16 OCOC Min Temperature at 2mK- 273.16 OCOC Directional wind at 10m (Zonal and Meridional) ms -1 Magnitude and x 0.748 Wind speed at 2m, ms -1 Downward Solar Radiation flux Wm -2 x 0.0864MJ/m 2 /day Specific Humidity at 2m Pressure at ground surface KgKg -1 Pa Universal gas law and saturation vapor pressure equations Relative humidity, % Hourly Precipitation rateKgm -2 s -1 Scaling theory and factor with duration Max half-hour rainfall, mm

7. Gumera: CFSR vs. Gauge Data 1

8. Gumera: CFSR vs. Gauge Data 2

9. Monthly mean Rainfall

10. Strict sense simple scaling property: the probability distribution of maximum rainfall depth is invariant of time scale (Burlando and Rosso, 1996) Wide sense simple scaling property: extends the scale- invariant property to quantiles and moments If the reference duration is 1hr, then = D Maximum Half-hour Rainfall

11. Maximum Half-hour Rainfall - 2

12. Maximum Half-hour Rainfall - 3

13. CFSR for SWAT Modeling - 1

14. CFSR for SWAT Modeling - 2

15. CFSR for Spatial Downscaling

16. CFSR for Water Fluxes Study (percentage of water fluxes relative to rainfall in wet season)

17. Conclusions High resolution reanalysis data had great potential to improve modeling of landscape processes CFSR data has comparable performance as gauged climate data for SWAT modeling in Ethiopian highlands The spatial pattern of CFSR data is useful for spatial downscaling and bias correction of GCM data The water fluxes of the CFSR data could be to study large-scale fluxes without doing cumbersome data assimilation

18. CFSR Data Description CFSR_Grids_Data Grids’ Center Location File: CFSR_Eth_Grids.xls CFSR Data File Names: GP1 2 _Eth.csv : the grid number in west- east direction : the grid number in north- south direction CFSR_SWAT_Inputs Grids’ Center Location Files: *.DBF’ and *.txt SWAT Input File Names: 1 2.txt : climate variables code (PC, TM, HM, SL and WN respectively for rainfall, temperature, relative humidity, solar radiation and wind speed) CFSR_WGEN: monthly statistics and grids’ center location