基于多源遥感数据的景观格局及预测研究

以TM、中巴资源卫星和环境与灾害监测预报小卫星等遥感影像为数据源,利用ENVI 4.7、ARCGIS 9.2、IDRISI 15等软件,研究了西安市辖区的景观特征与空间格局,预测了未来的景观变化,提出了景观格局预测的数据转化和多距离空间分析的精简步骤.结果表明:研究区的景观本底是一个由林地和耕地构成的复合景观基质,建设用地在研究时段内呈现持续增加趋势,且2004-2011年间的增加量高于2000-2004年间的增加量；林地面积略有降低,林地和草地总面积略呈增长趋势,水域和未利用地面积变化较小.研究时段内的景观破碎化程度在降低,林地景观的连通性增强了,耕地的降低了.各景观类型在所设定的最大预期研究尺度下均呈现显著的聚集空间格局；各年和各景观类型之间的聚集、随机和离散的临界阈值差别相对比较大；水域和未利用地的空间聚集强度明显高于耕地、林地、草地和城乡建设用地；耕地和草地空间分布存在一个异质性最大的特征尺度,且均出现了聚集分布、随机分布和离散分布3种分布格局,以2011年最为明显.利用景观指数法和多距离空间聚类分析方法研究景观格局特征的效果要比单一的景观指数法较理想.CA-Markov模型模拟的结果基本能够反映未来的景观格局状况.

Analysis and forecast of landscape pattern in Xi'an from 2000 to 2011

Landscape pattern analysis is an important topic of landscape ecology. The ultimate goal of landscape pattern analysis is to link spatial patterns of landscape with ecological processes, and detect status of processes using landscape pattern information. Landscape metrics can represent the spatial distribution of landscape and it has been used as a common tool in landscape pattern analysis. There are some inherent limitations of landscape metrics. Multi-distance spatial cluster analysis based on the Ripley''s k-function can fetch up the landscape metrics faults. Based on the image data of Land Resources Satellite TM (2000), China-Brazil Earth Resources Satellite (2004), and Environment Disaster Monitoring and Forecasting Small Satellite (2011), future landscape change were forecasted with bringing forward simplified process on data transform of landscape pattern change prediction and multi-distance spatial cluster analysis, via analyzing landscape characteristics and its spatial pattern in Xi''an city, using ENVI 4.7, ARCGIS 9.2, and IDRISI 15 in this study..Results showed that, tremendous landscape patterns changes have taken place in Xi''an during the past two decades. Complex landscape matrix consisted of forest land and cropland accounted for 67.98% of the total landscape in 2000 and 54.46% in 2011 in the study area. Built-up land area increased from 33213.91 hm² to 68380.79hm², and amount of increased area between 2000 and 2004 was more than that between 2004 and 2011. The area of forest land decreased insignificantly. The shift of water and unused land area was insignificant. Patch number in the class and landscape level decreased obviously, which showed that decreasing landscape fragmentation, enhancing connectivity of forest landscape, and reducing connectivity of farmland was taking place during the study interval. Landscape spatial pattern had a significant aggregation in the expected maximum distance (40 km). The critical thresholds discrepancy of clustered, random, and dispersed distribution between different landscape types in different years were relatively evident; the spatial aggregation levels of water and unused land were significantly higher than those of farmland, woodland, grassland, and built-up land; there were the largest characteristic scale of heterogeneous spatial distribution of aggregated distribution, random distribution, and dispersed distribution for cropland and grassland. The effect of landscape pattern characteristic study via coupling landscape index analysis and multi-distance spatial cluster analysis is better than that of using single landscape index analysis. The simulation of CA-Markov model could basically reflect the future landscape changes. The method and the simplified step of multi-distance spatial cluster analysis by ARCGIS and data conversion for landscape pattern forecasting by IDRISI were proposed in this paper.