Characterizing Forest Disturbance Mechanisms Using Vegetation Change Tracker Model

Continuous forest disturbance and recovery have long been recognized as critical processes modulating the carbon flux between the biosphere and the atmosphere, which play a major role in regional climate change investigations. With high spatial resolution, frequent revisits, objectivity, and mass of free data acquisition source, Landsat images make dense monitoring of forest disturbance and recovery possible. The yearly Landsat TM/ETM+ data acquired from 1987 to 2011 were used to monitor forest changes in Yaoluoping national nature reserve in the current work. The Landsat ecosystem disturbance adaptive processing system(LEDAPS) algorithm was applied to convert original Landsat DN values into surface reflectance, feeding into the vegetation change tracker model(VCT) to generate forest disturbance and recovery databases, followed by an intensive validation through on-screen visual comparisons of before- and after-disturbance images. On this foundation, the spatio-temporal characteristics of forest change, disturbance types, disturbance frequency and disturbance causality were analyzed. In addition, the spatially explicit processes of forest fragmentation were also characterized by implementing a forest fragmentation process model, and the pathways of disturbance influencing forest fragmentation process were obtained, to provide support for the strategic development of forest sustainable management. The main results were summarized as follows :(1)The VCT algorithm can effectively and accurately detect forest disturbance and recovery information in Yaoluoping national nature reserve. The Kappa coefficient of VCT products reaches to 0.79 with an overall accuracy of 80%, indicating a high applicability of VCT for forest disturbance monitoring at the landscape scale. Phenological issues and registration errors were identified as the main factors leading to a low precision of VCT products in certain years.(2) There was a fluctuating forest disturbance rate in Yaoluoping national nature reserve during the time period 1987 to 2011. The yearly disturbed area varied significantly, and the disturbance area during the period of 1988-1999 was larger than that of 2000-2011. Forest recovery area increased in the earlier years followed by a decrease in the latter stage, and the largest recovery area occurred in 1994. The forest cover rate fluctuated greatly, and reached the maximum of 91.25% in 2007. In space, with an exception that the disturbance area in the core area was larger than that in the buffer zone took place in 2009, which was largely different from the normal situation that the disturbance area in the core area was smallest and the experimental zone was largest in other years. The main disturbance types in Yaoluoping included logging, agricultural expansion and snow damage. During the entire study period, forest disturbance event occurred only once at any locations within the study site, without double disturbance events happened at the same location. The natural factors, socio-economic and political factors were the main driving forces responsible for the observed forest changes.(3) The forest fragmentation process model applied can capture four spatially explicit processes including perforation, subdivision, shrinkage, and attrition. From the view of quantity, forest loss patches were dominated by shrinkage type. From the perspective of area, shrinkage still played a dominant role. Four fragmentation processes both the quantity and area were all mainly distributed over the experimental zone, followed by the buffer zone, the core area rank the third. The contribution of human disturbance was greater than natural disturbance to the forest fragmentation. The main pathways of forest fragmentation were shrinkage and perforation, which resulted from logging and agricultural expansion.