【摘要】 目前我国土地资源面临着严重的盐碱退化问题。以上海市崇明岛东滩盐碱土为研究对象,基于野外实地调查土壤盐分数据以及Landsat遥感影像数据计算获取的各波段反射率、盐分指数(salinity index,SI)、盐分指数1(salinity index 1,SI1)、归一化差分植被指数(normalized difference vegetation index,NDVI)、冠层盐分响应指数(canopy response salinity index,CRSI)和陆地表面水分指数(land surface water index,LSWI),采用多元样条自回归模型(multivariate adaptive regression splines,MARS)与偏最小二乘回归方法(partial least squares regression,PLSR)分别建立土壤盐分的回归模型,并对区域盐碱土的空间格局进行探究。结果表明:(1)滨海土壤盐分在近红外波段有明显的吸收作用,与近红外波段、短波红外波段和NDVI相关系数较高;(2)MARS模型较PLSR模型对于样点土壤盐分反演有更好的效果(R2分别为0.74和0.70);(3)崇明东滩滨海土壤盐分在空间上具有较高的异质性,水体附近和滩涂土壤盐分较高,林地和农田土壤盐分较低。该结果为滨海地区区域尺度上的土壤盐碱化监测提供范例,为滨海土壤盐渍化的治理及岛屿的生态建设提供参考依据。更多还原

【Abstract】 Currently,the national land resources are facing a serious problem of salt alkali degradation. Taking Chongming Dongtan saline alkali soil in Shanghai as object,this study applied multivariate adaptive regression splines model( MARS) and partial least squares regression( PLSR) to establish regression models of soil salinity and explore spatial pattern of regional saline alkali soil,based on the filed sampling soil saline data,and band reflectance,salinity index( SI),salinity index 1( SI1),normalized difference vegetation index( NDVI),canopy response salinity index( CRSI) and land surface water index( LSWI) calculated from Landsat remote sensing data. The results showed that:(1) The coastal soil salinity performed obvious absorption in infrared band,and showed high correlations with infrared band,shortwave infrared band( SWIR1) and NDVI.(2) MARS model had better performance in retrieving of soil salinity than PLSR( R2= 0.74 and 0.70,respectively).(3) There was high spatial heterogeneity of soil salinity in Chongming Dongtan coastal area,with higher value near water body and intertidal zone,and lower value in forest and farmland. This paper present a fashion for the regional monitoring of soil salinization in coastal area,and providedvaluable information for controlling coastal saline alkali soil deterioration and ecological construction of the island.更多还原