Raindrop Size Distribution Parameters Dataset for Deep Learning Time Series Forecasting

DSD data comes from 20 OTT Parsivel² disdrometers deployed across Hainan Island, China. The dataset has 91 rainfall events from February 2019 to July 2023. This dataset is organized into three subfolders corresponding to the training set, validation set, and test set used in the deep learning model development. Each subset contains two subfolders: one for samples and one for labels. The samples and labels are stored as NumPy binary files (.npy), where each file can be loaded as an array with three columns and multiple rows. The three columns represent the min-max normalized values of the raindrop size distribution (DSD) parameters μ, lgNw, and Dm, respectively. If non-normalized data are required, the original parameter values can be recovered by applying the inverse min-max transformation using the normalization parameters provided in the associated publication.