Tracking land use trajectory to map abandoned farmland in mountainous area

Abstract

The strategy of recultivating abandoned farmland (AF) has been proposed as an effective response to the current global food crisis, it could increase food production without damaging the ecological environment. However, accurately identifying the spatial-temporal distribution of AF is a prerequisite for its successful implementation. Since abandonment is more likely to occur in mountainous areas than in plains, it is difficult to develop reliable methods to obtain long-time series information due to data source constraints and complex algorithms. In this study, we present a method to identify abandoned farmland based on tracking land use change trajectory in mountainous areas. Using Google Earth Engine (GEE), we mapped the land use classification of mountainous areas year by year and analyzed the land change at pixel level to obtain abandonment data through time series recursion. We applied this method to Tongjiang County, a mountainous area in China, and verified its accuracy, which turned out to be 82%. Our results indicate that the change in abandonment rate from 2001 to 2015 showed a phased characteristics that were likely determined by the interaction between policy, economics, and the rational choice of operators in different periods. Additionally, the Kernel Density Estimation (KDE) of AF distribution in Tongjiang County presented an agglomeration and stability pattern of “southwest> central> northeast”. Land transformation model (LTM) simulations further indicated that the future contraction or expansion of AF would have the greatest impact on the critical areas (southwest region of Tongjiang County). Our findings suggest that improving the precision of preferential agricultural policies, promoting the transfer of rural land management rights, and improving farming conditions in key areas could effectively address the problem of abandonment.