The leakiness index (LI) is an effective tool to measure the potential of landscapes in retaining resources (e.g., water and soil) that are vital for plant growth and environmental improvement in arid and semiarid landscapes by using vegetation coverage (VC) and digital elevation model (DEM) data derived from remote sensing images. However, sensitivity and applicability of LI in subtropical area have been rarely reported. In this study, effect of data resolution of the input parameters (VC and DEM) of LI was analyzed by using upscaling and downscaling methods. We also try to demonstrate application potential of LI in the Beijiang riparian vegetation buffer zone, a subtropical area, by comparing the result produced from LI with that from the universal soil loss equation (USLE). Results indicated that sensitivity of DEM to resolution and its influence on LI were lower compared with VC. LI sharply increased from 0.33 to 0.68 with decreasing VC, but minimally varied with increasing DEM. Performance of LI showed fluctuating changes in 1995, 2000, 2005, 2010, and 2015 with leakiness values of 0.09, 0.12, 0.21, 0.24, and 0.23, respectively. LI was significantly and positively correlated with erosion rate calculated using USLE in the corresponding period (p = 0.025), suggesting that LI can exhibit high potential in indicating soil and water conservation function in the targeted area. With decreasing data resolution (from 30 to 60, 80, 90, 120, 250, and 500 m), LI displayed a knee point at 120 m resolution scale; hence, 120 m resolution could be the critical scale for using LI to larger regions. This preliminary attempt can provide a better understanding of LI, and will contribute to the limited knowledge regarding the sensitivity and application potential of LI to subtropical area. [ABSTRACT FROM AUTHOR]