Method for identifying and quantifying atmospheric carbon monoxide plume based on multi-source satellite cooperation

Abstract

The present application provides a multi-source satellite cooperative-based atmospheric carbon monoxide plume identification and quantification method, comprising: training a ViT model using TROPOMI data to realize global CO plume intelligent rough screening, and generating a first-level mask by SAM after detecting an event; introducing GaoFen-5 AHSI data for a first-level event, performing matching filtering to invert a CO enhancement field and driving SAM to generate a second-level mask; introducing WorldView-3 data for a second-level event, inverting a CO enhancement field through a logarithmic radiation ratio and driving SAM to generate a third-level mask. Finally, according to the resolution of each level of satellite, the IME method is used to estimate the emission rate of each level of point source. The present application constructs a three-level cooperative automatic detection system of 'global scanning-regional detailed investigation-local precise investigation', effectively improves the automation level and attribution ability of CO point source detection, and is suitable for emission verification of key industries and facilities.

Images