Precipitation is a vital element of an atmosphere that has direct effect on Hydrology. Therefore, an accurate estimation of precipitation is obligatory to predict runoff through the catchment. Nepal has a high spatial variability of precipitation due to highly undulating surface terrain and complex relationships between land elevation and precipitation. This requires dense rain gauge network for the study of precipitation patterns in Nepal. However, it is difficult to arrange dense setup in the mountainous topography which is expensive and sometimes, even impossible for regular maintenance. The remedy for this problem could be the use of satellite based precipitation products with high temporal and spatial resolution.
In this study, TRMM data are employed to calculate rainfall over Nepal. TRMM rainfall product (that is precipitation data) can be obtained as: 3-hourly, daily, monthly and annually. These data are extracted and compared with 264 rain gauge stations in Nepal over the period of 8 years for their validation. A number of scripts are developed with python programming language to process raw satellite data which includes clipping to the region of interest, aggregation of day, month and annual values and extraction of rainfall on the gauging stations.
The day to day comparison showed that TRMM usually underestimates rainfall with some exception of overestimation at some regions. The CPOD_S value of 0.8 depicts that TRMM can detect precipitation in most of the days. Furthermore, in monthly comparison, TRMM estimated lower rainfall values but, the result still appeared to be acceptable (R² between 1 and 0.5) for use in other hydrological analysis. The comparison of annual precipitation as recorded by gauge and TRMM showed similar distribution patterns but the underestimation from monthly to yearly accumulate to form a huge difference in data sets. The correlation coefficient ranges from -0.86 to +0.8. The comparison of mean monthly precipitation data sets also derive same conclusion, but still possess a good R² efficiency. Though a fixed relation between altitude and intensity of rainfall could not be identified, it is observed that the efficiency of estimation decreases with increase in altitude.
The reliability of point to pixel (PO-PI) comparison method is also studied. This shows a wide spatial variability even within one pixel. The trend shows increasing precipitation on higher elevation. TRMM precipitation product estimates an average value falling over the pixel.
Since the comparison of monthly precipitation data showed good results, TRMM monthly precipitation data sets are also verified for use in discharge estimation from a catchment. A monthly water balance model developed by ‘Thornthwaite’ is simulated for four of the major river basins of Nepal including one sub-basin namely Karnali, Narayani, Bagmati, Saptakoshi and Sunkoshi. TRMM data without any bias correction factor is used for the estimation of discharge. The simulation results showed quiet an acceptable result although it is unable to catch up peak floods in some years.
Since TRMM is underestimating rainfall in most of the conditions, it will be wise to use an appropriate correction factor. Thus, a conclusion can be drawn from this study that TRMM rainfall products can be used in catchments with less or no rain gauge data, for estimating runoffs with a generalized bias adjustment factor.