岩溶区峰丛洼地山体阴影区域植被指数随机模拟

Sequential Simulation of Normal Different Vegetation Index of Mountain Shadow in Karst Peak Cluster Area

针对岩溶峰丛洼地遥感影像山体阴影区域信息缺失的问题,以广西壮族自治区果化野外基地的归一化植被指数(NDVI)为研究对象,对研究区域NDVI进行了14次序贯高斯模拟(SGS),并就模拟结果与普通克里格进行对比分析。结果表明,受岩性和地形地貌等内在因素的影响,研究区域NDVI具有强烈的空间自相关性;随着模拟次数的增加,SGS模拟值与验证值之间的相关系数增加,平均绝对误差(MAE)和均方根误差(RMSE)减小,模拟精度逐渐提高;当模拟次数大于50次时,模拟精度高于普通克里格(OK)插值精度,模拟次数继续增加,SGS模拟精度提高不明显。研究较精确地获取了阴影区域的NDVI,可为区域石漠化评估与岩溶生态重建提供参考。

Aiming at the problem of the missing shadow information of remote-sensing images existing in the karst peak and depression area, 14 relations of SGS （Sequential Gaussian simulation） for NDVI （Normal different vegetation index） of Guohua ecological experimental area were obtained in Guangxi Zhuang Autonomous Region. The simulation results were compared with ordinary Kriging. Validation results indicated that the NDVI of the study area was impacted intrinsic factors and performed strong spatial autocorrelation. With the increase of the number of simulations, the correlation coefficient between the simulation and verify values increased, and the mean absolute error （MAE） and root mean square error （RMSE） reduced. Along with the increase of simulation times, SGS simulation precision was gradually improved, and the precision of SGS was higher than that of OK （Ordinary Kriging） interpolation when simulation times were more than 50 times. Using SGS methods to predict the missing shadow information can provide a new idea and method for evaluation and ecological reconstruction of the karst rock desertification.