Rainfall estimation over Bangladesh using satellite data

Abstract

An attempt has been made to estimate rainfall in Bangladesh from GMS-5 data using Convective Stratiform Technique (CST) and these estimated rainfall was calibrated with the raingauge rainfall for all the Bangladesh Meteorological Department (BMD) raingauge stations located at different parts of the country. At first, rainfall and cloud characteristics were studied. It was found that the average lifetime of a single cell cloud was about 12 hours while multi-cell cloud lasted about 18 hours. The speeds of the movement of clouds were found to be higher during developing stages than the mature stages and the movement of cloud usually followed the direction from south or southwest to northeast or cast. The average correlation coefficient between raingauge rainfall and cloud top temperature was found -0.46. Due to the lack of the suitable image processing software fDr GMS-WEFAX data, a number of computer programs have been developed for image processing and for CST analysis. All these prohTfamswere written in 'C' language, compatible in all MS-DOS I Windows based operating system and can run independently wi th little execution time and effort. To adapt CST algorithm in Bangladesh region, some modifications were made on the calculating slope parameter, determining stratiform threshold temperature and assignment of both convective and stratiform rain rates. Due to the different data resolution for Bangladesh region, the slope parameter of each convective core, k should be 0.167 and it should be calculated from six surrounding pixels. The value of x to determine stratiform threshold temperature varies between the range 5K to 7K for all the raingauge stations of the country. Convective rain rate 20 mm/h and stratiform rain rate 3.5 mm/h were found best fit in this study. CST rainfall was calculated for the whole country and calibrated it with rainfall from all the BMO raingauge stations. It was found that CST rain/all had better agreement with raingauge rain/~lll than the previously calculated GMS-5 Precipitation Index (GPI) rainfall for Bangladesh. By CST method the deviation from raingauge was varied from 1.2 mm to 15.2 mm per day while by UPI method it was varied from 1.0 mm to 22.5 mm per which was over estimated. Though, results obtained from this study looks satisfactory and encouraging, rainfall estimation always need long term database for verification, it may be recommended to continue research put into that direction. In future efforts, it is suggested to use radar data if possible to verify the assigned convective and stratiform rain rates.