Modeling and Evaluating Systematic and Random Errors in Multiscale GPM IMERG Summer Precipitation Estimates Over the Sichuan Basin

Abstract

The Integrated Multi-satellitE Retrievals for the Global Precipitation Measurement (IMERG) is widely used in hydrological and meteorological studies, owing to its high spatial resolution and accuracy. The quantification of systematic and random errors for the IMERG estimates at different temporal resolutions is beneficial to the calibration of observation instruments and correction of estimates, especially for regions such as the Sichuan Basin where frequent geological disasters occur in the summer. At present, there are two most commonly used models, namely the additive and multiplicative models, for modeling the systematic and random errors of precipitation. However, it is unknown which model is more suitable for the IMERG summer hourly, daily, and monthly precipitation estimates. Therefore, in this study, two models’ separative capability of the systematic and random errors and predictive capability in the IMERG estimates are investigated, upon the evaluation of the models’ applicability. Results show that for the hourly and daily precipitation estimates, the multiplicative error model has better separative and predictive capabilities than the additive error model and is recommended to quantify the systematic and random errors. Conversely, as for the monthly precipitation estimates, the additive error model is a relatively better choice by comparing the overall performance of both models. However, it still has some weaknesses for heavy monthly precipitation, such as nonconstant fluctuations and reduced predictive capability. So, the additive model should be used with caution in analyzing the systematic and random errors of the heavy monthly precipitation.