Evaluation of the Effectiveness of Multiple Machine Learning Methods in Remote Sensing Quantitative Retrieval of Suspended Matter Concentrations: A Case Study of Nansi Lake in North China

Abstract

Total suspended matter (TSM) is a core parameter in the quantitative retrieval of ocean color remote sensing and an important indicator for evaluating the quality of the aquatic environment. This study selects part of Nansi Lake in North China as the study area. Researchers used Hyperion remote sensing data and field-measured TSM concentration as data sources. Firstly, the characteristic variables with high correlation were selected based on spectral analysis. Then, seven methods such as linear regression, BP neural network (BP), KNN, random forest (RF), and random forest based on genetic algorithm optimization (GA_RF) are used to construct the inversion model of TSM concentration. The retrieval accuracy of each model shows that the machine learning models are much more accurate than the linear model. Among them, the GA_RF model retrieves the suspended solids concentration with the best performance and the highest prediction accuracy, with a determination coefficient R2 of 0.98, a root mean square error (RMSE) of 1.715 mg/L, and an average relative error (ARE) of 6.83%. Additionally, the spatial distribution of TSM concentration was inversed by Hyperion remote sensing image. The results showed that the concentration of TSM was lower in the northwest and higher in the southeast, and the concentration distribution was uneven, showing the characteristics of a typical shallow macrophytic lake. This study provides an effective method for monitoring TSM concentration and other water quality parameters in the shallow macrophytic lake and further proves the advantages of machine learning in ocean color inversion. All in all, this research provides some useful methods and suggestions for quantitative inversion of TSM concentration in shallow macrophytic lakes.