Probable Maximum Precipitation (PMP) is an essential prerequisite in designing dams, spillways, and reservoirs in order to minimize the risk of overtopping infrastructure collapse, especially under today’s changing climate. This study investigates conventional PMP estimation approach by using both scarce in-situ observations and mainstream satellite precipitation products in the Dadu River basin, where plenty of reservoirs and dams are being built. The satellite data include Climate Prediction Center (CPC) MORPHing algorithm (CMORPH), Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks-Climate Data Record (PERSIANN-CDR), and Tropic Rainfall Measuring Mission (TRMM) Multi-satellite Precipitation Analysis (TMPA) 3B42V7. The evaluation of satellite products shows that CMORPH and 3B42V7 agree well with gauge-based dataset for the period of 1998–2013 at both the grid and basin scales, also capturing the extreme precipitation events, with high Correlation Coefficients (CC) in terms of 0.68 and 0.71, respectively. Also, CMORPH and 3B42V7 show better performance for the magnitude and spatial distribution of 24-h PMP in such complex terrains. PERSIANN-CDR shows an overestimation in the upstream and an underestimation in the downstream. As among the first studies of satellite precipitation-based PMP estimation, this work sheds lights on the suitability of satellite precipitation in PMP estimation and could provide a reference for future extended spatially-distributed PMP estimation in vast ungauged regions.
This study compares the advantages and limitations of traditional CMIP6 data fusion methods and machine learning fusion methods when applied to drought identification in the Yangtze River Basin. We consider three traditional fusion methods and five machine learning fusion methods, and calculate drought indices over 3-, 6-, and 12-month periods based on precipitation data from meteorological stations in the Yangtze River Basin (1960–2014) and 15 CMIP6 model datasets. The drought identification index is used to evaluate the performance of the fusion methods. Results indicate that traditional statistical methods have significant limitations in the upper reaches of the basin, where the terrain is highly undulating, but perform better in the middle and lower reaches, which are relatively flat. Among the machine learning methods, neural networks tend to amplify the observational noise, whereas kernel-tuning methods better accommodate nonlinear relationships across different SPI time scales. The prediction performance of all methods decreases from the 12- to 3-month drought indices, but the extent of the decline varies. The Random Forest and Radial Basis Function methods give the smallest reduction in performance, while the Backpropagation and Backpropagation-Adaboost methods produce the largest drop in performance.