Validation of a TRMM-based global Flood Detection System in Bangladesh

Although the TRMM-based Flood Detection System (FDS) has been in operation in near real-time since 2006, the flood 'detection' capability has been validated mostly against qualitative reports in news papers and other types of media. In this study, a more quantitative validation of the FDS over Bangladesh against in situ measurements is presented. Using measured stream flow and rainfall data, the study analyzed the flood detection capability from space for three very distinct river systems in Bangladesh: (1) Gangesa snowmelt-fed river regulated by upstream India, (2) Brahmaputra - a snow-fed river that is braided, and (3) Meghna - a rain-fed and relatively flashier river. The quantitative assessment showed that the effectiveness of the TRMM-based FDS can vary as a function of season and drainage basin characteristics. Overall, the study showed that the TRMM-based FDS has great potential for flood prone countries like Bangladesh that are faced with tremendous hurdles in transboundary flood management. The system had a high probability of detection overall, but produced increased false alarms during the monsoon period and in regulated basins (Ganges), undermining the credibility of the FDS flood warnings for these situations. For this reason, FDS users are cautioned to verify FDS estimates during the monsoon period and for regulated rivers before implementing flood management practices. Planned improvements by FDS developers involving physically-based hydrologic modeling should transform the system into a more accurate tool for near Hal-time decision making on flood management for ungauged river basins of the world. (c) 2010 Elsevier B.V. All rights reserved.