An Operational Real-Time and Seasonal Hydrologic Forecasting System (HFS) for the Upper Zambezi Basin, Africa

Improving the ability to generate temporally-continuous real-time or seasonal streamflow forecasts can potentially enhance the reliability of flood warnings, and improve water allocation and planning, among others, at basin-scale. However, the operational discontinuity of modelling efforts in the Upper Zambezi River Basin (UZRB) is still a significant challenge, which also represents a common problem in many basins around the world. As a way to overcome this issue, we have developed, implemented, and validated two operational Hydrologic Forecasting Systems (HFS) for the UZRB: (1) a Real-Time Hydrologic Forecasting System (RT-HFS), and (2) a Seasonal Hydrologic Forecasting System (S-HFS). The RT-HFS was setup for two time scales, short- and medium-range. In short-range, three Satellite Precipitation Products (TMPA-RT, CMORPH, and PERSIANN) are downloaded, bias-corrected, and used as forcings to three different distributed hydrologic models (HYMOD_DS, HBV_DS, and VIC 4.2.b), to produce daily (near) real-time streamflow forecasts for the basin. In medium-range, forcings are extracted from the Global Forecast System (GFS) outputs and fed to the three hydrologic models to produce daily streamflow forecasts for the next 8 days in the future. To forecast daily streamflows at longer time scales, the S-HFS was set up with hydrologic forcings from the North American Multi-Model Ensemble (NMME) project. Ten NMME ensembles of daily precipitation and temperature are downloaded in a monthly-basis to run the three hydrologic models to generate daily streamflow forecasts for the next 180 days in the future. All the data generated by both HFS are hosted on a web-based platform (SERVIR Water Africa-Arizona Team (SWAAT) website: www.swaat.arizona.edu), that provides streamflow forecasts at different time scales for multiple pilot basins in Africa. Present and future challenges in the implementation of both operational HFS are evaluated and discussed in terms of their advantages and disadvantages, with particular focus on analyzing the skill of the multi-product and multi-model approach, which forms the basis of the HFS.