基于遥感的草原退化人为因素影响趋势分析

 干旱、半干旱地区草原退化是我国重要的生态和经济问题,其通常是自然与人为因素综合影响的结果。而区分这两者的影响,无疑对草原退化的恢复重建工作具有重要的意义 。该文在基于前人的工作基础上,发展了一种分时段的回归分析和残差分析相结合的方法来评价人为因素在草地退化区域中的影响趋势。具体而言,以内蒙古自治区锡林郭勒草原为例 ,利用1983～1999年8 km空间分辨率的逐旬NOAA/AVHRR NDVI数据及同期气象数据为基础数据源,对每个像元建立1983～1988年的气候因子（考虑温度、降水,及其时滞效应）与 NDVI的回归关系模型,并逐年分析1989～1999年回归关系模型的预测 NDVI值与实际 NDVI的残差及变化趋势,从而判断以1983～1988年为基准各像元受人为因素影响的趋势。结果表明,除东乌珠穆沁旗以外,锡林郭勒草原的各旗（市）20世纪90年代都出现了较为严重的人类活动影响 加剧的草原退化,但退化的严重程度不一,呈现出东部和中部较轻、西部较重的空间格局,整个研究区域已达到中等退化程度。该结果与锡林郭勒草原退化的现状基本相吻合。

ANALYSIS OF HUMAN-INDUCED STEPPE DEGRADATION BASED ON REMOTE SENSING IN XILIN GOLE, INNER MONGOLIA, CHINA

Grassland degradation in arid and semi_arid steppe results from the integrated impacts of nature and human factors and is one of the most important ecological and economic issues in China. It is necessary to discriminate between natural and human_induced degradation for purposes of rehabilitation and restoration of degraded steppe regions. Based on existing research, we proposed a method that integrated regression analysis and residual analysis to discriminate between areas degraded by human activities and those caused naturally. We applied this method to the Xilin Gole Steppe in Inner Mongolia, China as a case study using Pathfinder NOAA/AVHRR NDVI data from 1983 to 1999 and meteorological data within the same time period. Firstly, we constructed a regression model using the first 6 years (1983_1988) of data for each pixel between annual maximal NDVI and meteorological data (precipitation and temperature) including a time lag for precipitation. Secondly, the difference or residuals between actual and predicted maximal NDVI were derived for the latter 11 years (1989-1999). The analysis of the trends in the residuals of the first data set over time indicated that pixels with negative trends were human_induced degradation. The results showed that the Xilin Gole Steppe experienced a heavier degree of human_induced degradation from 1989 to 1999 with larger areas being impacted. The results are in accord with the status quo of the study area.