Satellite observation completion method based on reanalysis data and unbalanced learning

Abstract

The invention discloses a satellite observation completion method based on reanalysis data and unbalanced learning, and the method comprises the steps of proposing an R2S framework, employing relatedvariables in the reanalysis data to simulate satellite observation, thereby filling the blank of satellite observation, constructing an R-S data set through employing an STM method under the R2S framework, and obtaining a satellite observation completion model suitable for tropical cyclone sea surface wind speed, wherein the R2S framework can significantly improve the space coverage rate and timeresolution of satellite observation; the invention further provides an SIMBA method, the performance of the complementation model at the high wind speed is improved through unbalanced learning, the method is combined with conventional machine learning to obtain a hybrid complementation model; the hybrid model is superior to the conventional machine learning model in the aspect of high wind speed complementation and superior to the unbalanced learning model in the aspect of medium-low wind speed complementation, and the completion result of the hybrid model is close to the field observation value, and the completion result is accurate.

Description

technical field [0001] The invention relates to the technical field of satellite observation, in particular to a satellite observation complement method based on reanalysis data and unbalanced learning. Background technique [0002] Satellite observation refers to the data obtained by low-orbit artificial earth satellites using remote sensing methods to observe and measure the earth's surface. Most scientific satellites and meteorological satellites operate in low orbits. Low-orbit satellite remote sensing images have high spatial resolution and shorter orbital periods. They can cover the whole world in a short period of time and have strong global observation capabilities. However, the observation swath of low-orbit satellites is limited, and they can only observe part of the earth's surface at a certain time, and cannot continuously observe the whole world. Therefore, in low-orbit satellite observations, the spatial coverage and temporal resolution are insufficient, and t...

AI Technical Summary

Problems solved by technology

[0008] The purpose of the present invention is to solve the problem of data gaps in most satellite observations in the prior art due to various factors such as the limitation of measuring equipment, the interference of the environment, and the delay or loss of data updates, and proposes a method based on reanalysis Satellite observation complementation method for data and unbalanced learning to improve the spatial coverage and temporal resolution of earth surface monitoring and improve the accuracy of satellite observation data through unbalanced learning

Images