Next Generation of a Real-time Global Flood Monitoring System Using Satellite-based Precipitation and a Land Surface Model

A real-time Global Flood Monitoring System (GFMS) using Tropical Rainfall Measuring Mission (TRMM) Multi-satellite Precipitation Analysis (TMPA) rainfall products and a hydrologic model is operating routinely and producing flood detection and intensity results (http://oas.gsfc.nasa.gov/globalflood/). This current system is discussed elsewhere in this session and was evaluated recently regarding the performance in flood detection against available flood event archives (Wu et al, 2012). This evaluation provided a pathway forward for improvement through the use of the hydrologic model component of the GFMS by coupling a more physically-based land surface model, the Variable Infiltration Capacity (VIC) Model, with a newly developed routing model. The VIC model also includes a snowmelt and soil frost process, which will also benefit the GFMS in forecasting spring streamflow and snowmelt related floods. The VIC model was adapted from its original individual grid cell based running mode to a mode that is suitable for real-time runoff prediction.

A hierarchical Dominant River Tracing-based Routing (DRTR) model was developed and coupled with the adapted VIC model for runoff calculation and routing to derive streamflow for all grids at each time step. The DRTR model simulates streamflow for each grid cell of the global domain by applying Kinematic wave or Muskingum-Cunge method hierarchically on dominant rivers coded with the Strahler river order, while the subgrid flow of each grid cell contributed from tributaries and overland areas within the cell is simulated using kinematic wave method. The DRTR model also outputs additional flood information such as flood magnitude, timing, flood wave celerity, river depth, potential inundation extent and depth etc. The DRTR model is run at 1/8th degree spatial resolution for global real-time flood detection, while a set of nested runs of DRTR model are triggered at higher resolution (e.g. 1km) for local areas where there are potential floods identified from our global detection algorithm.

The new VIC/DRTR based GFMS is initially evaluated based on streamflow gauge data and archived flood records and also verified against the performance of our old GFMS. Further evaluation and analysis are conducted based on intercomparisons of simulations by the VIC/DRTR model driven by different rainfall products (e.g. TMPA V6, TMPA V7, TMPA RT, Q2, NLDAS-2) for contiguous United States at variable temporal-spatial resolutions.