The main objective is to develop a seasonal water deficit forecasting system that is relevant for USAID and USACE activities in the Middle East and Africa based on existing NASA and NOAA Earth science capabilities. Our primary goals include: (1) supporting USAID’s Famine Early Warning Systems Network (FEWS NET) to help better predict water supply deficits related to agricultural drought and food insecurity, and (2) provide a suite of indicators related to forecasted water supply anomalies and conditions and of interest to our end-users.
Recent project highlights:
1) The key Land Surface Models (LSMs), Noah-MP, Catchment, Noah 3.3 and VIC, have all been run for 35-year historic simulations for East Africa. The results have been compared and are being evaluated for streamflow and drought metrics. Full water and energy budget analysis and streamflow evaluations have been performed.
2) In addition to the 35-year historic simulations, FLDAS Noah 3.3 model runs are being conducted in near-real time (with a latency of a couple days to 1 month). Outputs are being provided to NASA’s GES-DISC and are being used by the FEWS NET community.
3) A prototype for the Blue Nile River region has been set up to test and evaluate our full end-to-end forecasting system and perform a more complete analysis for all the components outlined. Land Information System (LIS)-based GEOS-5 ensemble forecasts have been run for several years and are being compared against climatological and Ensemble Streamflow Prediction (ESP) based runs. Bias-correction methods are in place for the GEOS-5 forecasts and will be tested and implemented before the end of the year.
4) The LIS-based HyMAP routing and stream flow model can now generate ensemble simulations, which supports the ensemble water availability and streamflow forecasts.
5) Spatial bias has been identified in some of the North American Multi-Model Ensemble (NMME) model forecasts for June-September precipitation in East Africa. This spatial bias results in low model skill under standard forecast evaluation techniques. However, a regionalization-based adjustment to account for spatial bias was found to enhance the skill of some NMME models. This points to a potential method for making use of predictions from spatially biased dynamical modeling systems.
6) Global, daily CHIRPS precipitation files have been temporally downscaled using the hourly MERRA-2 corrected precipitation field using the Land Data Toolkit (LDT) for the entire period of record and will be supported in near-real time (latency of 2-3 weeks).
7) Irrigation modules have been implemented and tested with most of the land surface schemes in LIS.
Geographic Focus
Africa and the Middle East
Principal Investigator
Dr. Christa Peters-Lidard (GSFC)
Project Team
Randy Koster (NASA/GSFC), Kristi Arsenault (SAIC, Inc.; NASA/GSFC), Sujay Kumar (NASA/GSFC), Bala Narapusetty (ESSIC; NASA/GSFC), Augusto Getirana (ESSIC; NASA/GSFC), Shrad Shukla (UCSB), Ben Zaitchik (JHU), James Verdin (USGS), Chris Funk (USGS/UCSB), Jeanne Roningen (DoD/CRREL), John Eylander (DoD/CRREL)
Collaborators and Stakeholders
Collaborators:
Amy McNally (ESSIC; NASA/GSFC), Hahn Jung (SSAI; NASA/GSFC), Fritz Policelli (NASA/GSFC), Greg Husak (UCSB)
Stakeholders:
Jim Verdin (USGS/USAID)
Rachael McDonnell (ICBA)
Technical Overview
The project team is working to develop an innovative system capable of harnessing available forecast skill and land state memory/initial conditions. These factors, which previous work has demonstrated to be essential for effective drought early warning, are a major part to our overall forecasting system. Existing tools and datasets are being used and include: 1) the LIS framework; 2) skilled GEOS-5 and CFS seasonal forecasts; and 3) land state initialization via assimilation of AMSR-E/ASCAT/SMOS/SMAP soil moisture and GRACE terrestrial water storage.
The first phase of the project has focused on augmenting the LIS-based FLDAS, consisting of WRSI, Noah, and VIC LSMs with 1) models that include a prognostic water table (the Catchment and Noah-MP LSMs); and 2) with properly downscaled seasonal forecasts from NASA’s GEOS-5 and NOAA’s CFS. Downscaling and bias-correction of the seasonal forecasts are being implemented to improve the existing forecasts and overall drought and water supply anomaly forecasts. In addition, we are working to improve the initial conditions of the seasonal forecasts using satellite-based observations of the land surface which are assimilated into the LSMs, like using NASA’s GRACE TWS products and NASA’s SMAP soil moisture mission. To evaluate the system, skill and information content are being assessed for the major components of the system, separately and integrated, to provide our end-users information on the overall performance and benefit of the system.
The second phase is an integral part of implementing this work with regional USAID and
USACE partners, such as ICBA in UAE. These partners will help to evaluate and demonstrate
the forecasting system, including the use of downscaled water deficit hindcasts and forecasts for
crop production estimates. ICBA will support engagement of the in-country user community to
enhance their existing decision-making based on our water deficit forecasts. This will be
accomplished with standard FEWS NET procedures and methodologies relating to projected crop stress to yield, and use of state-of-the-science crop models such as DSSAT, to explicitly
model the water deficit impact on crop growth. These estimates can be evaluated using yield data
as well as market/census data.
Additional Information
Earth observations / models / technologies applied:
NASA’s LIS framework LIS website)
NASA’s MERRA-2 reanalysis and GEOS-5
NOAA’s Climate Forecast System (CFS) seasonal forecasts
Land surface models: Noah LSMs, NASA’s Catchment LSM, and the DSSAT crop model
Observed precipitation: RFE2 and CHIRPS satellite-insitu precipitation data, GPM/TRMM precipitation
Other relevant remotely sensed products: GRACE TWS, MODIS and VIIRS based vegetation, land temperature and ET datasets, AMSR-e/SSMI/SMAP/ASCAT soil moisture products