Water environment monitoring method based on multi-source remote sensing and machine learning

Abstract

The invention relates to the technical field of water quality monitoring, in particular to a water environment monitoring method based on multi-source remote sensing and machine learning, which comprises the following steps: preprocessing data of a monitoring station for researching historical water quality of a water body; processing, analyzing and fusing the multi-source historical remote sensing image of the research area to obtain multi-source historical remote sensing data; carrying out space-time fusion on the actual water quality parameters of the monitored water body and the corresponding multi-source historical remote sensing data to form a water quality inversion data set, and preprocessing the water quality inversion data set; inverting water quality parameters by using a machine learning algorithm, and preferentially selecting a water quality inversion model through performance evaluation; according to the method, compared with the prior art, real-time monitoring and early warning of water environment pollution events such as red tide in a wide area range of an offshore area are achieved, the space-time precision of remote sensing data monitoring is effectively improved, and the method has the advantages that the real-time monitoring and early warning of the water environment pollution events such as the red tide in the wide area range of the offshore area are achieved. Environmental adaptability and practicability of the model are enhanced, and offshore water environment pollution control is supported.

Description

technical field [0001] The invention relates to the technical field of water quality monitoring, in particular to a water environment monitoring method based on multi-source remote sensing and machine learning. Background technique [0002] The near-coastal sea area refers to the sea area adjacent to the mainland coast, islands, and archipelago, and the outer limit of the territorial sea is on the landward side. The sea area under the jurisdiction of China is about 3 million square kilometers, the coastline is more than 18,000 kilometers, and the coastal land area is less than 5% of the country, but the population and GDP account for 22.2% and 36.8% of the country, and the industrial and domestic pollution loads are extremely high. According to 2017 According to the annual investigation results of land-based pollution sources entering the sea, among the 9,600 land-based pollution sources in the country, there are more than 740 rivers entering the sea and more than 7,500 sewa...

AI Technical Summary

Problems solved by technology

However, the processing method of satellite remote sensing data in the existing technology has a high error rate, low utilization rate of satellite remote sensing data, weak temporal and spatial accuracy of water quality monitoring, and prolonged time, so real-time monitoring cannot be achieved; the application of machine learning algorithms in the existing technology cannot effectively fit the water quality time sequence, it is difficult to capture the potential relationship between the reflectance of remote sensing images at any location and the measured water quality data through a specific algorithm, and the adaptability to different environments is poor

Images