Search

A seamless and effective early warning system is critical for mitigating the most adverse impacts of water and food insecurity in Africa. The U.S. Agency for International Development (USAID) supported Famine Early Warning Systems Network (FEWS NET) is tasked with providing early warning and analysis on acute food and water insecurity to help government decision-makers and relief agencies plan for and respond to resulting humanitarian crises. This presentation describes a recently developed multimodel based seasonal scale hydrologic forecasting system, which has been operationally supporting the early warning efforts of FEWS NET since August 2018. Since then, this system has helped identify flood risk potential in parts of Eastern and Western Africa, and development and progression of drought conditions in Southern Africa. Both of these types of extremes contribute to food and water insecurity. At the core of this system is NASA's Land Information System (LIS) (a hydrologic modeling framework), which supports seasonal scale monitoring and forecasts of runoff and soil moisture by combining satellite and reanalysis based atmospheric forcings datasets and dynamical climate forecasts. Specifically, the LIS-based seasonal scale hydrologic forecasting system incorporates NASA/GMAO's Catchment and the Noah Multi-Physics (Noah-MP) land surface models (LSMs). It is driven by NASA/GMAO's Modern-Era Retrospective analysis for Research and Applications, version 2 (MERRA-2), and the UCSB Climate Hazards Group InfraRed Precipitation with Station (CHIRPS) rainfall dataset to provide a long historic record for monitoring (1982-present) of hydrologic variables. Currently, the seasonal scale hydrological forecasts are generated by using NASA's Goddard Earth Observing System Model, version 5 (GEOS-5).Long-term availability of monitoring and forecasts of hydrologic variables allows for historical validation of this dataset against independent hydrologic observations and reported crop yields (up to the extent they are available). Validation results thus far have demonstrated agreement of hydrologic monitoring simulations with independent satellite based observations in Eastern Africa and the value of this system in supporting food insecurity early warning in Southern Africa.This presentation will first describe the development of this system, followed by its validation for drought early warning and subsequently the process of stakeholder engagement and operational application of the products derived from this system by the FEWS NET. [ABSTRACT FROM AUTHOR]

• A hydrologic forecasting framework is built to predict drought and flood potentials for food insecurity early warning. • Hydrologic forecasts are improved by using entire suite of NMME total-precipitation forecasts in land surface modeling. • The higher skill of the entire suite of NMME precipitation forecast is also reflected in predicting hydrologic drought. In situ hydrologic monitoring over regions most susceptible to food insecurity can be a challenge in current times due to various socio-economic and political issues in combination with environmental factors such as ongoing famine or drought. Hydrologic monitoring and initializing forecasts based on remotely sensed and analyzed data can contribute significantly to early warning in such regions. Routine hydrologic forecasts, as provided by NASA's Hydrologic Forecasting and Analysis System (NHyFAS), are a recent addition to early warning systems. A custom instance of NHyFAS, termed FLDAS-Forecast, is used by FEWS NET's Land Data Assimilation System (FLDAS). The FLDAS-Forecast's dynamic forecasting component was originally set up with Goddard Earth Observing System (GEOS) forecast inputs and has been recently expanded with precipitation forecast forcing from the North American Multi-Model Ensemble (NMME). This paper describes the improvements in seasonal hydrologic forecasts produced with this updated system. Evaluations in this study focus on soil moisture across southern Africa's growing season. Soil moisture forecasts are benchmarked and evaluated relative to climatology-based forecasts and historic runs, which are driven by observation-based meteorological forcing fields, and they are verified with remotely sensed observations of soil moisture and vegetation. Through multiple deterministic and probabilistic skill assessments, we show that using the larger ensemble of NMME precipitation inputs in the forecast system results in higher quality hydrologic forecasts than are allowed by climatology- or GEOS-only-based forecasts. Further, the near-real-time NMME-based rootzone soil moisture forecasts were able to correctly predict developing drought conditions over southern Africa through late 2019 and into early 2020. [ABSTRACT FROM AUTHOR]