A Comprehensive Analysis of Pixel-level Rainfall Datasets for the Indian Region: Identification of Optimal Rainfall Datasets

The objective of this study is to identify the rainfall datasets for India that deliver the best results at the pixel level. Rain is water droplets that have condensed from atmospheric water vapor and then fall under gravity. Rainfall is a dynamic process and is constantly changing in form and intensity as it passes over a given area. The Indian Meteorological Department's (IMD) gridded data are utilized in this work to undertake skill metrics analysis on seven commonly used rainfall datasets, including GPM, CRU, CHIRPS, GLDAS, PERSIANN-CDR, SM2RAIN, and TerraClimate. The rule-based decision tree techniques are employed on the obtained skill metrics analysis values to find the good-performing rainfall dataset at each pixel value among all the datasets used. The temporal analysis (in both month- and year-wise scales) suggests that GPM is a good-performing dataset. This analysis identified the good-performing datasets in 3428-pixel locations out of 4641-pixel locations in India. The pixel-wise analysis reveals that GPM correlated well with the IMD dataset in 3105 pixels out of 4641 pixels, whereas TerraClimate correlated well only 1579 pixels. Among the seven chosen datasets, this analysis found the best dataset for each pixel. PERSIANN-CDR, followed by CHIRPS, and finally the GPM dataset, often scores as a good-performing fit. The TerraClimate dataset is the least valuable at the pixel level despite having a better resolution. For hydrologic and agricultural applications that support sustainable development, this research will help in the selection of the best dataset at a pixel of India.