Taking Ebinur Lake Wetland Nature Reserve as the study area, and based on the classified result of Landsat TM images of 1998, 2006, and Landsat OLI images of 2014, this paper simulated and forecasted land use/cover types of Ebinur Lake Wetland Nature Reserve by using the CA-Markov model. In the model-building process, the transition probability matrix and the transition area matrix were obtained through the Markov model, which determined the conversion rules and iterative times of the CA model. The land use/cover pattern of the study area in 2014, 2022 and 2030 was simulated and forecasted with the CA-Markov model. Then the forecast result was compared with the actual classified data of 2014 to verify the forecast accuracy. The prediction error of land use/cover type area was not more than 6.4% in 2014, and the spatial location accuracy was 76.0%. The results showed that the forest and grassland, saline-alkali area, dry lake bed and desert had an increasing trend in 1998-2014. The saline-alkali area increased the most by 37.4% and the water body decreased by 34.8%. In 2014-2030, the areas of the forest and grassland, saline-alkali soil and desert would increase, while that of the dry lake bed, water body and other objects would decrease. This study could provide a basis for dynamic monitoring land use/cover as well as sustainable development of Ebinur Lake Wetland Nature Reserve.
以艾比湖湿地自然保护区为研究区域,以1998、2006年Landsat TM影像和2014年Landsat OLI影像的土地利用/覆被分类结果为输入数据,采用CA-Markov模型,预测研究区未来的土地利用/覆被格局.在模型建立过程中,通过Markov模型求出转移概率矩阵和转移面积矩阵,确定CA模型转换规则,限制CA模型迭代次数.利用CA-Markov模型模拟预测研究区2014、2022和2030年土地利用/覆被格局,并采用2014年实际土地利用/覆被分类结果验证预测精度,得到2014年各土地利用/覆被类型面积预测误差均≤6.4%,空间位置预测精度达到76.0%.结果表明: 1998—2014年,艾比湖湿地自然保护区林草地、盐碱地、干涸湖床和沙漠增加,其中,盐碱地的增幅最突出,增加了37.4%;水体和其他地类减少,且水体的减少突出,减少了34.8%.2014—2030年,艾比湖湿地自然保护区林草地、盐碱地和沙漠将呈增加趋势,而干涸湖床、水体和其他地类将减少.该研究可以为艾比湖自然保护区的土地利用/覆被动态监测以及可持续发展提供依据.
Keywords: CA-Markov model; Ebinur Lake Wetland Nature Reserve; land use/cover; remote sensing; simulation and forecast.