An integrated approach for risk assessment of rangeland degradation: A case study in Burqin County, Xinjiang, China. (June 2020)

• Record Type: Journal Article
• Authors: Chen, Yan; Wang, Wei; Guan, Yang; Liu, Fangzheng; Zhang, Yubo; Du, Jinhong; Feng, Chunting; Zhou, Yue
• Is Part Of: Ecological indicators. Volume 113(2020)
• Publication Date: 2020-06
• Subjects: Rangeland degradation -- Risk analysis -- Rangeland vegetation coverage -- Random forest -- Bayesian belief network; Environmental monitoring -- Periodicals; Environmental management -- Periodicals; Environmental impact analysis -- Periodicals; Environmental risk assessment -- Periodicals; Sustainable development -- Periodicals; 333.71405
• DOI: 10.1016/j.ecolind.2020.106203
• Languages: English
• ISSNs: 1470-160X

Abstract

Highlights: Trend analysis, driving force identification and risk prediction were carried out. Random Forest Regression was applied to identify driving force. Climatic variables appeared to be important drivers of vegetation coverage dynamics. The reliability and efficiency of risk assessment were enhanced by Bayesian Belief Network. Current enclosed rangelands poorly represented the area with the highest risk. Abstract: Rangeland degradation in China has significant impacts both on the ecosystem and on the pastoralists' life. However, the emphasis of current management practice is always put on the rangeland with serious degradation problem. How to effectively avoid the degradation risks is still unclear. Thus, an integrated approach for rangeland degradation risk assessment was designed, consisting of the analysis of vegetation dynamics, driving forces identification and degradation risk prediction. Firstly, Vegetation Indexes and field survey data were applied to build regression model to calculate the vegetation coverage status and trend of change in each grid. Secondly, the important driving forces of rangeland dynamics were identified based on the local knowledge and objective data. Thirdly, Bayesian Belief Network (BBN) was trained in each seasonal rangeland to predict the probability of rangeland degradation in each grid. Rangeland degradation in Burqin County, Xinjiang, China was served as a case to test the practicability of this approach. The results indicated that during 2000 and 2013 most of the rangeland grids remained stable. Twelve factors were identified to be the driving force of the trend of vegetation coverage dynamics. BBNs showed that in most of the study area degradation risk was less than 50%, and the grids facing the maximum risk were only appeared in a small range. According to the case study, the integrated approach based on Random Forest and BBN model was turned out to be a practical and effective tool for the risk assessment of rangeland degradation.