Flood Simulation With Distributed Hydrological Approach Using Dems And Remotely Sensed Data

Jinkang Du,Shunping Xie,Youpeng Xu,Hua Xie,Yujun Hu,Peifa Wang,Shunfu Hu
DOI: https://doi.org/10.1109/IGARSS.2006.272
2006-01-01
Abstract:The spatial variability of both meteorological and geographic features has a great influence on rainfall runoff process. When making flood forecasting, the influence of this variability on runoff should be considered. In this paper, the distributed hydrological modeling technique was adopted to simulate flood discharge with consideration of spatial variability using DEMs and remotely sensed data. In our approach, the grid based digital elevation data was used for the presentation of watershed discretization and deriving flow path system and parameters. The infiltration and rainfall excess on each hillslope grid cell were calculated using the Green-Ampt infiltration equation. Soil lateral flow was estimated based on Darcy's law and continuity equation, overland and channel flows were calculated by using one-dimensional kinematic wave equations on hillslope and channel grid cells respectively. The land use and land cover data derived from remotely sensed images was used for the determination of Manning's roughness coefficients. The soil type data was used for deriving soil hydraulic parameters needed for calculation.The approach was applied in Jiaokou Watershed with area of 259km(2), a sub-basin of Yongjiang River in Zhejiang Province, China. The data collected for this study included land cover processed from Landsat TM images, soil type distribution from the soil maps, 100-m spatial resolution DEMs produced from digital topographic maps. Spatial distribution of rainfall for each flood event was calculated by using distance inverse interpolation methods. Nine flood events were simulated, two of which were used for calibration, and the others for validation. The simulated hydrographs at basin outlet showed good agreement with observed ones for both calibrated and validated flood events, and average coefficients of efficiency were 94% for calibrated flood events and 92% for validated flood events.
What problem does this paper attempt to address?