Eco-Environmental Quality Monitoring in Beijing, China, Using an RSEI-Based Approach Combined With Random Forest Algorithms

Abstract

The assessment of ecological environmental quality (EEQ) has provided an important knowledge base for protecting human health and realizing sustainable development. Previous studies have often used only principal component analysis (PCA) to perform the EEQ evaluation by determining the remote sensing based ecological index (RSEI) in a single year, and the assessment results are not comparable between years. Thus, a comparable and accurate method needs to be found and applied. In this paper, we applied the PCA combined with a random forest algorithm (a machine learning algorithm) to quantify the EEQ of Beijing, China, in 2014 and 2020 and analysed the relationship between the RSEI and four ecological indicators (greenness, wetness, dryness and heat). The results suggested that the RSEI and the ecological indicators of Beijing all changed substantially from 2014 to 2020, and the method of combining PCA and random forest was suitable for calculating the time-series data of RSEI in the study period. Specifically, the RSEI in Beijing increased slightly from 0.31 to 0.33 overall, the greenness of Beijing increased drastically (26.09%), the wetness decreased by 10.00%, and the dryness and heat increased by 8.62% and 2.00%, respectively. The Pearson correlation coefficient test showed that both the greenness and wetness had positive effects on the RSEI, while the dryness and heat had negative effects. Of the four ecological indicators in Beijing, the greenness contributed greatly as the main positive factor, and dryness was the most negative factor during the six years. This paper developed an improved framework for continuous EEQ monitoring, and these results provide a scientific basis for the sustainable development and ecological environmental monitoring of Beijing and other megacities.