All-time three-dimensional atmospheric pollutant reconstruction method and device, computer equipment and storage medium

Abstract

The invention discloses an all-time atmospheric pollutant reconstruction method and device, computer equipment and a storage medium, and the method comprises the steps: obtaining the atmospheric pollutant concentration distribution obtained through the inversion of hyper-spectral remote sensing data of an observation region, and obtaining the multi-source related information of the atmospheric pollutant concentration in the observation region, the multi-source related information comprises meteorological information, traffic information and topographic information, the multi-source related information is used as a sample after being preprocessed, and atmospheric pollutant concentration distribution is used as a sample label; constructing a multi-modal reconstruction model comprising a feature extraction unit and a reconstruction unit; carrying out supervised learning on the multi-modal reconstruction model by utilizing the samples and the sample labels so as to optimize parameters of the multi-modal reconstruction model; and reconstructing all-time three-dimensional atmospheric pollutant distribution by using the parameter-optimized multi-modal reconstruction model. According to the method, the full-time three-dimensional atmospheric pollutant concentration is comprehensively predicted by considering the multi-modal information, so that the prediction accuracy is improved.

Description

technical field [0001] The invention belongs to the field of air pollutant concentration monitoring, and in particular relates to a full-time air pollutant reconstruction method, device, computer equipment and storage medium. Background technique [0002] For air pollution monitoring, the generation, traceability and diffusion prediction of pollutants are very important. A single point instrument or vertical profile observation instrument can only monitor and analyze the pollutant discharge in the block where it is located, but cannot trace the source of trace gases from other blocks, and cannot identify the source of pollution. Therefore, with the goal of predicting and reconstructing the three-dimensional air pollutants in the whole region, it is urgent to build a joint prevention and control list of regional emissions to monitor, prevent and trace the 24-hour pollutants in all blocks. [0003] In the field of atmospheric pollutant detection, the existing mainstream detec...

AI Technical Summary

Problems solved by technology

[0003] In the field of air pollutant detection, the existing mainstream detection equipment is mainly divided into three types. The first one is based on point instruments such as CNEMC (National Environmental Air Quality Monitoring Station), which cannot detect air above the ground or at high altitudes. Contaminants are detected

The second is radar, which can detect atmospheric pollutants at different heights, but radar instruments have blind spots

This prediction method directly considers the regional spatio-temporal distribution characteristics of pollution source pollutants, that is, considers meteorological information, pollutant information and time stamp information to make predictions, which will cause relatively inaccurate predictions

Images