AN INTER-COMPARISON OF SATELLITE BASED NOAA CPC RAINFALL ESTIMATES AND GAUGE OBSERVATIONS OVER SELECTED STATIONS IN INDIA

The spatial application of crop simulation models need grid based spatially distributed input of rainfall. The global products of satellite based rainfall estimation can provide spatial maps on regular time basis and can be used as an input to crop models. An attempt has been made to validate the satellite derived NOAA CPC rainfall estimation with ground based measurements for year 2003 and 2004 over 53 stations of India Meteorological Department (IMD) in India on daily, 5-day (pentad), weekly, 10-day (dekadal), monthly and seasonal scales. The mean bias between NOAA CPC estimated rainfall and IMD observed rainfall were 3.0, 3.9, 5.8, 10.6 and 50.5 mm over mean observed rainfall of 19.8, 31.2, 40.0, 128.8 and 1106.8 mm for pentad, weekly, dekadal, monthly and seasonal totals, respectively. The satellite estimates showed Willmott’s index of agreement ranged from 0.91 to 0.97 and correlation coefficient from 0.84 to 0.94 for pentad to seasonal composites with the measured rainfall in plain areas. The percentage mean bias reduced from 15.2 to 4.5 per cent as period of comparison increased from pentad to season. The mean bias is below 5 mm for rainfall estimates upto 100 mm while it increases upto 25 –30 mm for rainfall estimates from 100-200 mm and above. For weekly estimates, the mean absolute error and root mean square error observed with 1:1 line were 15.0 mm and 30.2 mm, respectively. The estimates are very close to seasonal totals of observed rainfall for most of the stations except stations having altitudes greater than 500 m