Abstract: Multi-source precipitation products play important roles in estimation of possible maximum precipitation (PMP), especially for data-sparse regions. This study evaluates the accuracy of three global precipitation products－i.e. Climate Hazards Group Infrared Precipitation with Station (CHIRPS), Multi-Source Weighted-Ensemble Precipitation (MSWEP), and Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks (PERSIANN)－based on the Asian Precipitation-Highly-Resolved Observational Data Integration Towards Evaluation (APHRODITE) dataset and in-situ measurements in the Lancang-Mekong River basin. An improved statistical method is used to estimate the 1d and 3d PMPs. Results show that (1) the MSWEP dataset presents the satisfactory accuracy and precipitation detection ability; (2) the MSWEP-based PMP is significantly correlated with the APHRODITE results at the grid-cell scale; (3) the MSWEP-based 1 d and 3 d PMP estimates are 122 and 192 mm, respectively, comparable to the previous estimation results. This study indicates that the global precipitation products are useful to PMP estimation for data-sparse regions.

Key words: multi-source data, probable maximum precipitation, statistical method, design flood, Lancang-Mekong River basin