Forest Disturbance Detection By Different Remotely Senseddata

Forest disturbance directly reflects the people and the interaction of forest ecosystem and forest succession processtion; it should not be neglected for the influence of the ecological environment. Remote sensing detect land vegetation change and study the basis of the carbon cycle, for regional or larger scale, the remote sensing technology is and effective means of continuously monitoring the forest change on a regular basis. Jiangxi wuning county as study area, this paper used30m resolution of TM/ETM+data from1986to2011;30m resolution HJ satellite data from2009to2011; MODIS09A1of8days surface reflectivity products from2004to2012;250m resolution of FY normalization vegetation index products from2010to2012, selected vegetation index and disturbance index time series method, respectively, compared the forest disturbance information extraction ability. By the different spatial resolution and time resolution remote sensing data, determine the best way to extract forest disturbance information, and the best data for forest disturbance monitoring. Combined with elevation, water system, road conditions in the study area, analysis the law of forest disturbance, and study the drive factors of the forest disturbance. Research mainly divided into the following several parts:1) The HJ-CCD data and TM/ETM+data, for the same four band correlations are good. Using TM data from1986to2011selected the DI (Disturbance Index) time series reconstruction method to distill information of forest Disturbance. At the same time will be2009/2011of the HJ-CCD and2009/2011of the TM/ETM+and HJ-CCD data set by using NDVI (normalized difference vegetation index) threshold extraction forestland and two image NDVI difference threshold to extract forest disturbance information. TM compared with HJ-CCD forest disturbance information extraction precision by difference of NDVI value threshold way is higher, at67.2%and78.9%.For HJ-CCD, the difference of NDVI difference threshold method to extract the result accuracy is higher, at69.3%. TM/ETM+_DI index extraction results is the best, the accuracy is86.67%. So extract method and wavelength range to extract the disturbance information is more important.2) The vegetation index through the S-G (Savitzky-Golay) reconstructed. The results of the S-G eliminates the clouds and atmospheric effects and clear description of time series of long-term trends. Extraction disturbance imformation show that after reconstruction of MODIS NDVI disturbance information extraction is optimal in the studied area, the accuracy reached76.76%by the RCI (turbulence intensity index) method. FY data affected by cloud, accuracy is only45%.So for the same method.The data the spatial resolution is not entirely decided extraction ability of disturbance information, data selection influence the result of forest disturbance extraction.3) Using MODIS with Tasseled Cap matrix calculation MODIS image DI index.Set DI threshold to determine the best accuracy is78.3%, compared with MODIS RCI and TM/ETM+DI, TM/ETM+DI at best accuracy was86.67%. Results show that compared with traditional vegetation index time series, disturbance index reduces data redundancy. Affected by the spatial resolution, MODIS can’t detects the small area forest disturbance information, when forest disturbance occurs with larger area, using MODIS NDVI and DI method can detect the forest disturbance information. Compared with RCI method, MODIS-DI for forest disturbance information is more sensitive.4) In order to study the law of forest disturbance and its driving factors, we selected Wuning in Jiangxi Province as study area. We have adopted method of remote sensing index time series trajectory analysis to determine the disturbance region using14Landsat TM/ETM images during1986-2011and analyzed the change rule of forest disturbance and determined the driving factors with the EOF (Empirical Orthogonal Function) method. Wuning County suffered from the most dramatic disturbance in the1990s, especially in1992, and the trend of disturbance was going down from1986to1998. In contract, forest recovery was going upward from1992to2011. Logging and forest fire are the main disturbance factors and forest recovered relatively faster. The disturbance mainly occurred in nearly road, low altitude and near water areas. The EOF analysis shows that Qingjiang, Henglu and Shidu present three forest disturbance centers and Xinning has the lowest degree of disturbance. The forest disturbance mainly affected by forest management policy as well as the terrain and traffic restrict them. Meanwhile, urbanization will gradually play an important driver factors. The conclusion can provide important reference for forest disturbance studies in South China.