Framework for rainfall estimation using X-Band polarimetric radar

Abstract

Several countries, including Japan and the United States, have developed and researched X-Band polarimetric radar technology for disaster preparedness applications. However, this weather radar is still unavailable in tropical countries such as Malaysia, which has experienced flash floods and flood events, particularly during the monsoon season. The ability of X-Band polarimetric radar to observe rainfall for a localised area (30-60 km of observation radius) using a dual polarisation method can provide detailed and precise information to the disaster area in one or two minutes. Therefore, this research aimed to develop the framework of rainfall estimation by using X-Band polarimetric radar data. This research used Design Science Research (DSR) approach as the methodology. First, a systematic literature review was conducted to identify the current weather radar monitoring system in Malaysia and examine the theoretical background of X-Band polarimetric radar and its applications, as well as the components of the X-band polarimetric radar framework. Second, the DPPC server was set up with the study area and rainfall data of Kagoshima City. The study area data comprises Digital Elevation Model (DEM) and drainage area; whereas the rainfall data comprises X-Band polarimetric radar and AMeDAS ground data. Third, the framework components - storage, analysis tools and visualisation – were analysed using the X-Band polarimetric radar data to process the actual radar data and estimate the rainfall in four selected drainage areas. The analysis includes preprocessing raw data, clipping the DEM, selecting the drainage area, extracting rainfall data, and plotting the hyetograph. This is followed by the reliability analysis of the rainfall estimation from the extraction of X-Band polarimetric rainfall data analysed based on AMeDAS ground data. The validation results showed that there is a moderate positive correlation between X-Band polarimetric and AMeDAS with R2 of 0.24 and p = 0.009. Further, the comparison of rainfall distribution results for both data by using QGIS showed that AMeDAS data was not reflecting actual rainfall monitored by XBand polarimetric radar due to no AMeDAS stations in the drainage area and the distribution image from X-Band polarimetric radar was in real-time. All four drainage areas recorded heavy rain in the upper stream and showed the exact locations that had the highest intensity and the probability that the area closest to the drainage area would flood. In conclusion, it has been demonstrated that the newly developed framework translates real raw X-Band polarimetric radar data into accessible engineering data formats and that it can be used to estimate rainfall in a localised area.