Reducing spatial resolution increased net primary productivity prediction of terrestrial ecosystems: A Random Forest approach

Zhou, Tao; Hou, Yuting; Yang, Zhihan; Laffitte, Benjamin; Luo, Ke; Luo, Xinrui; Liú, Dan; Tang, Xiaolu

doi:10.1016/j.scitotenv.2023.165134

Cited by 10 publications

(1 citation statement)

References 73 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It is highly resistant to overfitting, noise anomalies, and multicollinearity between variables [63,64], and is particularly good at dealing with nonlinear relationships [65]. Consequently, random forest algorithms have been widely used in regression prediction problems [66] and feature classification [67] in the ecological field. In addition, the random forest model can effectively assess and rank the importance of each variable [68].…”

Section: Introductionmentioning

confidence: 99%

Quantifying the Impact and Importance of Natural, Economic, and Mining Activities on Environmental Quality Using the PIE-Engine Cloud Platform: A Case Study of Seven Typical Mining Cities in China

Zeng,

Dai,

et al. 2024

Sustainability

View full text Add to dashboard Cite

The environmental quality of a mining city has a direct impact on regional sustainable development and has become a key indicator for assessing the effectiveness of national environmental policies. However, against the backdrop of accelerated urbanization, increased demand for resource development, and the promotion of the concept of ecological civilization, mining cities are faced with the major challenge of balancing economic development and ecological environmental protection. This study aims to deeply investigate the spatial and temporal variations of environmental quality and its driving mechanisms of mineral resource-based cities. This study utilizes the wide coverage and multitemporal capabilities of MODIS optical and thermal infrared remote sensing data. It innovatively develops the remote sensing ecological index (RSEI) algorithm on the PIE-Engine cloud platform to quickly obtain the RSEI, which reflects the quality of the ecological environment. The spatial and temporal evolution characteristics of the environmental quality in seven typical mining cities in China from 2001 to 2022 were analyzed. Combined with the vector mine surface data, the spatial and temporal variability of the impacts of mining activities on the ecological environment were quantitatively separated and explored. In particular, the characteristics of mining cities were taken into account by creating buffer zones and zoning statistics to analyze the response relationship between RSEI and these factors, including the distance to the mining area and the percentage of the mining area. In addition, the drivers and impacts of RSEI in 2019 were analyzed through Pearson correlation coefficients pixel by pixel with 10 factors, including natural, economic, and mining. Regression modeling of RSEI in 2019 was performed using the random forest (RF) model, and these drivers were ranked in order of importance through random forest factor importance assessment. The results showed that (1) the ecological quality of mining cities changed significantly during the study period, and the negative impacts of mining activities on the ecological environment were significant. (2) The areas with low RSEI values were closely related to the mining areas and cities. (3) The RSEI in the mining areas of mining cities was generally lower than the average level of the cities. The RSEI gradually increased as the distance to the mine site increased. (4) The increase in the size of the mine area initially exacerbates the impact on the ecological environment, but the impact is weakened beyond a certain threshold. (5) The distance to the mining area is the most important factor affecting the quality of the ecological environment, followed by DEM, GDP, and precipitation. This study is of great importance for advancing sustainable development in mining cities and formulating sustainable strategies.

show abstract