Atmospheric-hydrological modeling of severe precipitation and floods in the Huaihe River Basin, China

SummaryOur study focuses on the simulation of heavy precipitation and floods over the Huaihe River Basin (270,000 km2), one of the seven major river basins in China. The simulation covers two periods in 1998 (June 28–July 3, July 28–August 17) and a third period in 2003 (June 26–July 22). The former two periods, with eight meteorological cases each of duration 72-h, correspond to the Intensive Observation Period of HUBEX/MAGE (Huaihe River Basin Experiment/Monsoon Asian GEWEX Experiment). The period in 2003 with 10 cases is the second most severe flooding event on record. The Canadian atmospheric Mesoscale Compressible Community Model (MC2) is used for precipitation simulation in the hindcast mode for all cases. The Chinese Xinanjiang hydrological model driven by either rain gauge or MC2 precipitation is used to simulate hydrographs at the outlet of the Shiguanhe sub-basin (5930 km2), part of the Huaihe River Basin. The MC2 precipitation is also evaluated using observations from rain gauges. Over the Huaihe River Basin, MC2 generally overestimates the basin-averaged precipitation. Three of the eight 1998 cases have a percentage error less than 50% with the fourth having an error of 54%, while six of the ten 2003 cases have errors less than 50%. The precipitation over five different sub-regions and the Shiguanhe sub-basin of the Huaihe River Basin from MC2 are also compared with values from the Chinese operational weather prediction model; the latter data are only available for the ten 2003 cases. An excellent result is obtained in the hydrological simulation using rain gauge precipitation as revealed by the Nash–Sutcliffe coefficients of 0.91 for both summers of 1998 and 2003. The simulation using MC2 precipitation shows a reasonable agreement of flood timing and peak discharges with Nash–Sutcliffe coefficients of 0.63 and 0.87 for the two 1998 periods, and 0.60 for 2003. The encouraging results demonstrate the potential of using mesoscale model precipitation for flood forecast, which provides a longer lead time compared to traditional methods such as those based on rain gauges, statistical forecast or radar nowcasts.