شبيه سازي و هشدار سيل با تلفيق مدل‌هاي آبشناس در GIS و برآورد بارش از طريق سنجش از دور

Simulation and Flood Warning with Hydrology Models in GIS and Precipitation Estimation through Remote Sensing

Flood warning and forecasting could be one of the most effective non-structural procedures in managing floods in order to decrease the risks and disasters floods caused. The main aim of this paper is to investigate the application of RS and GIS techniques associated with hydrological model in relation to flood forecasting. To achieve the aim, Madarsoo river basin as a flood prone region in Golestan Province was selected. Because rainfall in August 10, 2005 caused flooding in the region, the images of NOAA/AVHRR satellite for this date were chosen. In the study, first, all layers required in GIS were made on the basis of the factors causing flood. Second, spatial database of parameters including stream, cross sections, direction of runoff, banks and ModClark grid precipitation model were created and placed into the hydrologic model. In order to identify and classify clouds and estimate the rainfall data provided by NOAA/AVHRR, object-oriented classification and cloud indexing method were used respectively. Then, the quantity of runoff, caused by the rainfall, was estimated by considering the soil and land use maps and other inputs data in hydrologic model. Finally, to create the flood map, according to the topographical characteristics of the region, two most effective factors namely the depth and speed of water were taken from the hydrologic model to the GIS environment again. The results of the research indicated that we can achieve acceptable accuracy by using object-oriented classification, so that the overall accuracy of classification in this investigation is about 0.905 and Kappa coefficient is about 0.887. It was found that Cumuluscongestus and Stratocumu-lus clouds with the rainfall rate of 10.8 mm/h and 2.2 mm/h respectively, had the most and the least contribution to making floods. According to the hydrograph driven from the middle and lower parts of basins and also assuming that our estimation is done from the beginning of the rainfall, the flood map and the time in which the flood occurred could be predicted 20 and 33 hours before to it gets peak, respectively.