基于Landsat 8遥感影像反演干旱区净初级生产力的尺度效应  

Scale effect of net primary productivity based on remote sensing in arid region

作　　者：杨会巾 刘丽娟[1] 马金龙[1] 王进 李小玉[1] 

机构地区：[1]中国科学院新疆生态与地理研究所,乌鲁木齐830011 [2]乌兰乌苏农业气象试验站,新疆石河子832003

出　　处：《生态学杂志》2016年第5期1294-1300,共7页Chinese Journal of Ecology

基　　金：国家自然科学基金项目(31470708;41271202;U1503182;41501205);荒漠绿洲国家重点实验室自主课题(Y471163)资助

摘　　要：遥感影像是不同尺度景观分析的数据源,不同尺度上占主导地位的地物景观信息是不同的;描述或解释许多环境问题时,都不能在单一尺度解决,因此多尺度分析已成为景观生态领域中的必要条件。本研究以玛纳斯河流域为研究区,以2013年分辨率为30 m的Landsat 8遥感影像作为原始数据,对其分别进行聚合平均,得到不同分辨率(30、150、330、660、990、1650、1980和3300 m)的遥感数据,并作为NPP遥感估算模型的输入参数,探究不同分辨率的植被净初级生产力情况。结果表明:随着尺度的扩展,农田净初级生产力均值随之下降,林地、草地和裸地净初级生产力估算值基本保持不变,就整体而言,分辨率30 m的植被NPP模拟值比分辨率990 m的NPP模拟值多5.36%。随着尺度的扩展,玛纳斯河流域净初级生产力平均值发生变化,产生尺度效应,主要是由土地覆盖类型引起的;而不同尺度的玛纳斯河流域NPP具有相似的空间格局,且其基本特征不变。Remote sensing is one of widely used data source of landscape analysis with different resolutions. Predominant ground object could be different on various scales. Multi-scale analysis has become a necessary method in the field of landscape ecology because a single scale could not describe or explain many environmental problems. The data, which were aggregated by average with different resolutions （30, 150, 330, 660, 990, 1650, 1980 and 3300 m） based on the Landsat 8 remote sensing images with original 30 m resolution of 2013, were input to remote sensing model to simulate multi-scale NPP. The results showed that, with scaling up, NPP of crop increased, but NPP of forest, grassland and bare land almost remained unchanged. Average stimulated NPP of 30 m resolution was 5.36% higher than that of 990 m resolution. Upscaling in- duced the change of mean estimated NPP in Mans River Watershed, producing scale effect, due to land cover change. However, the spatial patterns of NPP in Mans River Watershed were simi- lar on various scales and their characteristics remained unchanged with scaling up.

关 键 词：净初级生产力 尺度扩展 尺度效应 空间异质性 

分 类 号：Q948[生物学—植物学]