基于地质统计学影像纹理的海南矿区荒漠化监测
Desertification monitoring for Hainan diggings based on geostatistical texture
-
摘要: 多年来,由于对钛矿的无序开采,使得海南岛东部出现大面积的土地荒漠化.采用遥感的手段进行跟踪监测,合理地授予采矿权,组织适当的复垦,是解决当地荒漠化的有效途径.基于不同沙地类型在地表空间结构上的差异,提出将基于地质统计学的影像纹理应用到荒漠化监测中,通过变异函数纹理来加大各种不同类别沙地间的区别,提高样本选择的分离度.结果表明,运用变异函数纹理结合光谱波段的最大似然分类方法能够很好地界定海滩沙地和内陆荒漠地的等级,最高分类精度达到92.4%,证明了基于地质统计学的影像纹理在实现该地区遥感荒漠化监测方面的有效性.Abstract: There is great desertification in the east of Hainan Island of China due to over-mining of ilmenites. The effective methods to combat desertification are monitoring the change of land with remote sensing, licensing the rights of mining ilmenites rationally, and organizing the moderate reclaim. Based on the difference of sandy land types on the spatial constructions, geostatistical texture was used to monitor desertification, and the discrimination degree of sample selection could be increased by using variogram texture to increase the difference of different kinds of sandy land. The results show that the maximum likelihood classification based on variogram texture and spectral bands can perfectly define the grades of beach sandy land and inner desertification, and the maximal classification precision comes up to 92.4%, which proves that geostatistical texture is effective in the application of desertification monitoring.