Abstract:The identification of farmland at risk of abandonment is not only the scientific basis for utilization planning and management of overall farmland in a region but also a key measurement of agricultural sustainability. However, the accuracy of identification results can be affected by the gradualness and complexity of abandonment process and limited by the development of current relevant technologies. In response to these challenges, a set of scientific identification methods have been established for use in the World Heritage Site of Hani Rice Terraces. In the first of two steps, a farmland abandonment risk index (FARI) is constructed using factors such as road access, slope, elevation, water, precipitation, temperature, population, rice production, among others. Second, the farmlands at high risk of abandonment are identified through hierarchical cluster analysis. The results suggest first that farmland can be divided into three categories of abandonment threat:stable, low-risk abandonment, and high-risk abandonment. These categories accounted for 57%, 35%, and 8% of the total farmland area of the heritage site, respectively. Overall, the results indicate that farmlands at the heritage site were stable. Second, the variation trend for the FARI in longitude and latitude was:north was higher than south and east was higher than west. Third, in terms of spatial patterns, farmlands at high risk of abandonment (34.95 km ²) were distributed at the boundary of Yuanyang County and composed of two parts (from Ganiang and Fengchunling to Xiaoxinjie, from Majie to eastern Niujiaozhai). The region was a key area for ecological restoration in the region because of poor cultivating conditions. Fourth, by analyzing the factors influencing the three types of farmland, the following results were obtained. Stable farmland was primarily affected by population, rice production, and road access, followed by slope, water, and elevation. Stability was positively associated with population and rice production and negatively associated with road access, slope, water, and elevation. Farmland at low risk of abandonment was most affected by precipitation, and other factors had less impact. Farmland at high risk of abandonment was primarily affected by temperature, slope, elevation, and river connectivity. Among them, the risk was positively associated with slope, elevation, and river connectivity and negatively associated with temperature. In conclusion, the Hani Rice Terraces were generally stable, with little risk of abandonment, and the factors affecting the risk of abandoned farming differed significantly in space.