基于遥感与BPNN-CA模型的草场保护区模拟--以锡林浩特温带典型草原为例

Simulation of Grassland Protection Areas Using Remote Sensing and BPNN-CA Model:A Case of Xilingol Temperate Grassland

过度放牧等人类干扰活动是目前干旱、半干旱区草原退化的重要原因之一。建立草场保护区,对过度放牧等人类行为进行合理的管理和控制,是防止草原退化,改善草原生态环境的一项重要措施。利用遥感影像,土地利用/覆盖类型图以及相关数据,提出了一种遥感技术和BPNN-CA模型技术相结合建立草场保护区的新方法,并在中国内蒙古锡林浩特草原进行了实例研究。该方法包括:①通过遥感技术监测研究区草原现状,提取覆盖度高,质量好的区域作为种子区;②利用BPNN-CA模型,根据政府规划草场保护区总量要求,模拟建立草场保护区。结果表明,本文提出的方法能够快速有效的提取草场保护区,在快速监测草原状况的基础上,能够同时兼顾草场内在适宜性和空间分布合理性,模拟出不同规划要求下的草原保护区空间分布格局,相对于传统人工作业方法更加节约人力和物力,能够有效地为政府部门的规划决策提供科学依据。

Grasslands are significant natural resources that cover 41% of the global land surface. However, grasslands in arid and semi-arid regions are facing desertification or degradation caused by human activities and climate change. Grassland degradation could have a significant impact on the carbon cycle, regional economy and climate. Research has shown that grassland degradation has become a serious environmental problem in China due to climate variability and human disturbances in arid and semi＇arid areas. Establishing grassland protection areas （GPAs） is regarded as an effective measure to reduce degradation. The GPAs should be carefully zoned so that they can be effectively protected with reasonable investment. At present, a GPA is often zoned based on in situ investigation, which is subjective and timeconsuming. For a vast grassland region such as the Xilingol steppe grassland in northern China, it is virtually impossible to effectively manage and monitor the resources using field investigation methods. This paper presents a new method to support local governments in the zoning of GPAs. It integrates a cellular automata （CA） model with back-propagation neural network （BPNN）, geographical information system （GIS） and remote sensing. The proposed method includes three major steps： 1 ） Extract “seed points” of candidate GPAs using remote sensing techniques and GIS; 2） Determine the weights of factors used in the CA model using the BPNN model; 3） Simulate the zoning of GPAs using a BPNN-CA model. This method was applied to a case study in Xilingol steppe grassland in the Inner Mongolia Autonomous Region of China. A Landsat-7 ETM ＋ image acquired on May 23, 2000 was used for the case study, along with ancillary data including topographic maps, land use, transportation and soil maps at the scale of 1 ： 100, 000 obtained from the local government. All the data layers were registered to the same UTM coordinate system and sampled to a pixel resolution of 150 m. Six factors were included in the BPNN-CA model： slope, land use, neighborhood effect, distance to rivers, distance to roads and distance to urban areas. The results show that this integrated approach can be used to rapidly identify candidate GPAs that satisfy the zoning requirement. The advantage of this approach lies in its innovative implementation of zoning requirements in a BPNN-CA model and the utilization of advanced remote sensing techniques. For the case study in Xilingol steppe grassland, more than 95 % of the simulated candidate GPAs were actually located in high-quality grasslands with limited human disturbance. In addition, the simulated candidate GPAs are more spatially concentrated and therefore easier to manage and protect than the segmented results from the vegetation coverage map. The simulation of GPAs would therefore provide good candidates for local government selection of protection areas. This study demonstrated the potential applications of this approach in grassland protection, as a valuable decision support tool for government and planning agencies.