THE APPLICATION OF SATELLITE BASED PRECIPITAION FOR FLOOD INUNDATION MAPPING OF MIDDLE INDUS BASIN

Abstract

The impacts of floods and droughts are intensified by climate change, lack of preparedness and coordination. Pakistan lies in an arid and semi-arid climate zone with average rainfall ranging from less than 200 mm to more than 400 mm per year. In Pakistan rainfall is highly variable in its magnitude, time of occurrence and its spatial distribution. Rain gauge generally provides very accurate measurement of point rain rates and the amounts of rainfall but due to scarcity of the gauge locations provides very general information of the area on regional scale. Additionally, systematic and human errors make gauge data more prone to errors and mis-calculations. Recognizing these practical shortcomings of rain gauges data, it is essential to use remote sensing techniques for measuring the quantity of rainfall in the Middle Indus. This research uses satellite remote sensing of NASA’s Tropical Rainfall Measuring Mission 3B42V7 estimates in terms of rainfall occurrence, quantity and its spatial distribution for Middle Indus Basin from Chashma barrage to Guddu barrage to model flood inundation. In order to use TRMM satellite data for flood inundation mapping, its accuracy is determined by statistically comparing it with in-situ gauged data on daily and monthly basis. Daily comparison results mention the average statistics with CC= 0.64, RMSE= 4.69, Relative Bias%= -14 and R2= 0.42 and monthly comparison with average CC= 0.91, RMSE= 4.07, Relative Bias%= 26 and R2= 0.82. The daily R2 value is significantly lower than monthly R2 value, probably due to the time of summation of TRMM 3-hourly precipitation data into daily estimates as opposed to the rain gauge data. It is also possible that the TRMM estimates fail to record the local development phenomena common in monsoon season. The in-situ rain gauge data xiv are generally recorded at about 0300 GMT, whereas TRMM daily estimates are the summed up from of 3-hourly collections from 0000 hours to 2100 hours GMT. Hence a difference of at least 6 hours is encountered in summing daily rain estimates. Daily TRMM data from 2003 to 2012 was used as input forcing in Soil and Water Assessment Tool (SWAT) hydrological model along with other input parameters like Land cover, temperature and soil data etc. The calibration and validation results of SWAT model give R2=0.72 and 0.73, NSCE= 0.69 and 0.65 respectively. Daily flood inundation maps are generated on the basis of model discharge output. The results of this research can be used by various stake holders, such as Pakistan Meteorological Department, Federal Flood Commission, Provincial Irrigation Departments and other planning and development authorities to investigate, plane, and analyze of climate change, floods, droughts and weather conditions.