Establishing forest resilience indicators in the hilly red soil region of southern China from vegetation greenness and landscape metrics using dense Landsat time series

Abstract

Resilience is the capacity of an ecosystem to respond and recover from damage or stress. It can inherently exhibit the cycle and feedback of the disturbed ecosystem recovery process for a specified period. The current availability of dense and consistent time series of satellite images holds the promise of monitoring forest resilience. The aim of this study is to establish forest resilience indicators using dense Landsat time series and assess forest resilience in response to the maximum disturbance magnitude in the hilly red soil region, Hengyang Basin, Southern China. To achieve this, Landsat images of the study area from 1987 to 2017 were collected. Normalized Difference Vegetation Index (NDVI), i.e., proxy of forest green characteristics, number of patch (NP), average patch size (PS), patch perimeter-area ratio (PPAR) and aggregation index (AI), i.e., proxies of forest landscape metrics were calculated from Landsat images. And then elasticity (i.e., the time and rate of recovery), malleability (i.e., degree of deviation from an initial state) and trend (i.e., the pattern of change) as the indicators of forest resilience were constructed. The local space–time Moran’s I (STI) based on NDVI residual space–time Moran’s I (STI) was employed to characterize the forest disturbance and recovery process. The results revealed the following. Firstly, the STI calculated from dense time series NDVI residuals are successful at monitoring the forest disturbance recovery process, regardless of whether changes were dramatic or subtle. Secondly, the most forest disturbances (i.e., > 75%) occurred in the late 1980s and early 1990s. NDVI and landscape metrics also differed in their response to disturbances; NDVI, PPAR and AI are more malleable to disturbance than PS and NP are. Finally, approximately 40% of the disturbed forest had the strong elasticity with a short recovery time (i.e., a year). We conclude that measuring forest resilience via vegetation greenness and landscape metrics using dense time series satellite images is practical as an operational tool for policy makers, landowners, and national park managers. Moreover, insights into ecological dynamics are emerging from capturing the process showing the difference both before and after the change.