Evaluating intensity-duration-frequency (IDF) curves of satellite-based precipitation datasets in Peninsular Malaysia

In recent years the use of remotely sensed precipitation products in hydrological studies has become increasingly common. The capability of the products in producing rainfall intensity-duration-frequency (IDF) relationships, however, has not been examined in any great detail. The performance of four remote-sensing-based gridded rainfall data processing algorithms (GSMaP_NRT, GSMaP_GC, PERSIANN and TRMM_3B42V7) was evaluated to determine the ability to generate reliable IDF curves. The work was undertaken in Peninsular Malaysia. The best-fitted probability distribution functions (PDFs) of rainfall totals for different durations were used to generate the IDF curves. The accuracy of the gridded IDF curves was evaluated by comparing observed versus estimated IDF curves at 80 locations. The results revealed that a generalized extreme value (GEV) distribution had the best fit to the rainfall intensity for different durations at 62% of the stations, and this was then used to develop the IDF curves. A comparison of these remote sensing derived IDF curves with the observed IDF data revealed that the GSMaP_GC product performed best. In general, the satellite-based precipitation products tended to underestimate the IDF curves. The GSMaP_GC IDF curves were found to be the least biased (8%–27%) compared to the TRMM_3B42V7 IDF curves (65%–67%). The biases in rainfall intensity of different return periods for GSMaP_GC for all grid points were estimated. These results can be used in designing hydraulic structures where gauged data are unavailable. •Performance of remote-sensing-based rainfall data in IDF curve generation is evaluated.•The accuracy of remote-sensing IDF curves was evaluated using observed IDF curves at 80 locations.•Satellite-based precipitation products were found to underestimate IDF curves.•Biase in remote sensing rainfall data is corrected for developing reliable IDF curve.