The invention discloses a regional environment anomaly monitoring method based on
remote sensing inversion and
deep learning, and relates to the technical field of environment monitoring application, and the method is characterized in that multi-source
remote sensing data is integrated, three types of core environment data of
atmosphere, water and soil are inverted after preprocessing, monitoring units are divided in a unified manner, and a global
data matrix is constructed; three types of environment association feedback nodes are defined, an exception
list is generated by calculating an environment exception potential index, and the traditional data scattering limitation is broken through. According to the method, high-priority monitoring units are screened, a'
pollution source-
pollution factor-sensitive node 'full-chain monitoring association link is constructed based on three rules, risk indexes are quantified to mark abnormities, resources are concentrated, and the monitoring accuracy is improved. Meanwhile, unprocessed units are marked as dynamic adjustment units, correction indexes are calculated in combination with
multiple factors, the monitoring range is dynamically optimized, a closed-loop
system is formed, static monitoring defects are avoided, and monitoring timeliness and perspectiveness are improved.