Monitoring and analysis of abandoned cropland in the Karst Plateau of eastern Yunnan, China based on Landsat time series images

Abstract

The Karst Plateau in eastern Yunnan, China (KPEYC) is densely populated. In this area, rocks are widely exposed, and soil erosion and rock desertification are serious issues. Furthermore, the conflict between humans and land is prominent. It is important to monitor abandoned croplands to ensure regional food security, preserve the ecological environment, and promote sustainable regional development. Regarding the lack of historical data and inaccurate identification of abandoned cropland, this study proposed a monitoring method for abandoned croplands based on Landsat time series remote sensing images. First, a decision tree was used to generate constant samples from 2000 to 2020. Thereafter, land use in KPEYC was classified year-by-year based on the random forest method and abandoned cropland pixels were identified based on land use data. Finally, the spatial and temporal characteristics of abandoned cropland in the KPEYC were analysed. Based on the results, the overall accuracy of the method was 0.75. Cropland abandonment in the eastern Yunnan Karst Plateau displayed a rapid growth trend and a slow decrease during the study period, with a significant growth rate from 2001 to 2010. The abandoned cropland was mainly concentrated in southwestern Zhaotong City, the entire territory of Qujing City, and northwestern Wenshan Prefecture, among which Fuyuan County, Luoping County, and Huize County repeatedly experienced more serious cropland abandonment. In addition to social factors, topography had a greater influence on cropland abandonment in the KPEYC, and the greater the slope or altitude, the more severe the cropland abandonment.

Keywords

• Landsat time series
• Cropland abandonment
• Remote sensing classification
• Karst Plateau