A Correction Method of Satellite Rainfall Data Based on Machine Learning to Discriminate Rainfall Events

Abstract

The invention discloses a satellite rainfall data correction method based on machine learning to distinguish rainfall events, which belongs to the technical field of satellite rainfall data accuracy correction. The ability of rainfall data to capture the presence or absence of rainfall and improve the accuracy of satellite rainfall products. The method includes: 1) multi-source data collection, 2) data processing, 3) rainfall event discrimination model construction, 4) rainfall correction, 5) rainfall sequence segmentation, 6) distribution optimization, 7) rainfall data segmentation deviation Correction. The present invention first adopts a machine learning method to construct a rainfall event discrimination model to correct the presence or absence of rainfall of satellite rainfall data, and then adopts segmental appropriate distribution to correct the total amount of satellite rainfall, and then obtains higher-precision continuous satellite rainfall data. This method effectively improves the false alarm rate and missing alarm rate of satellite rainfall data, and significantly improves the accuracy of satellite rainfall data. Furthermore, optimally obtaining the appropriate distribution of rainfall sequences based on multiple distributions avoids the subjectivity of artificially given distributions, and can better eliminate errors between satellite rainfall data and measured rainfall.

Description

technical field [0001] The invention relates to a method for correcting satellite rainfall data based on machine learning for distinguishing rainfall events, and belongs to the technical field of precision correction of satellite rainfall data. Background technique [0002] Rainfall is one of the key links in the hydrological cycle, and its accuracy directly affects the simulation accuracy of runoff process and the design of flood control and flood control projects. Accurate rainfall data play a crucial role in studying the response of hydrological processes in the context of climate change and in modeling runoff processes in undatad areas. With the intensification of global climate change, the temporal and spatial distribution of rainfall has been greatly affected. How to simulate rainfall and improve the accuracy of existing rainfall products has gradually become an issue that requires extensive attention in the fields of atmospheric science and geography. [0003] In rec...

AI Technical Summary

Problems solved by technology

[0004] Although there are many methods for correcting rainfall deviation, most of the existing methods perform deviation correction on a certain aspect of rainfall, and do not consider much about the judgment and correction of whether a rainfall event occurs, which leads to the satellite rainfall’s ability to capture whether there is rain or not. Insufficient, high false positive rate and false negative rate

At the same time, in the existing application of distribution fitting to correct rainfall data, most methods use artificially given distribution for correction, so the given distribution will have the characteristics of subjectivity and time invariance, which is not conducive to the deviation correction of satellite rainfall in changing environments

Images