A methodology to upscale IMD ground radar observations at the same resolution with TRMM PR reflectivity using ANN

Abstract

This work presents a direct comparison of Ground Radar (GR) located in Delhi against Space Radar (SR) onboard the Tropical Rainfall Measuring Mission (TRMM). At first, the GR observations made by the India Meteorological Department (IMD) for the Indian summer monsoon months of 2013 over Delhi are matched with SR observations using a three-dimensional volume matching methodology (VMM). We found that the matched GR and SR reflectivity data points at common volumes have a high correlation coefficient with a high RMS error. Further, due to the beam broadening of the GR, the matched reflectivity has the exact horizontal resolution of 5 km but a different vertical resolution based on the distance from the GR site and the elevation angle. An upscaling methodology is proposed to obtain GR observations at an equal resolution (5km�0.25km) as that of SR. We observe a better agreement between GR and SR data following our approach. Further, the high vertical resolution GR reflectivity is �corrected� for bias and other sources of error using SR observations with artificial neural networks (ANN) based approach. Our results show that the ANN-corrected high-resolution GR reflectivity observations have a root mean squared (RMS) error of 3.98 dBZ, which reduced significantly from 16.35 dBZ. � 2023 Elsevier B.V.