基于光谱转换的土壤盐分反演与动态分析  

Soil salinity retrieval and dynamic analysis based on spectral band inter-calibration

作　　者：陈文娇 翁永玲[1] 范兴旺[2] 曹一茹 

机构地区：[1]东南大学交通学院,南京210096 [2]中国科学院南京地理与湖泊研究所,南京210008

出　　处：《东南大学学报（自然科学版）》2017年第6期1233-1238,共6页Journal of Southeast University：Natural Science Edition

基　　金：国家自然科学基金资助项目(41471352)

摘　　要：为了对黄河三角洲地区进行大范围、长期的盐渍土监测,以研究区2000—2016年间4景Landsat-5 TM,EO-1 ALI,Landsat-8 OLI时间序列影像及Hyperion高光谱数据为基础开展土壤盐分定量反演分析.将Hyperion数据按照光谱响应函数分别重采样为TM,ALI,OLI模拟数据,采用数值回归方法计算TM,ALI与OLI对应波段间的光谱转换系数,从而将TM,ALI影像转换为OLI时序影像.分别采用偏最小二乘回归模型与多元线性回归模型建立土壤光谱与盐分参量间的预测关系,并将最优预测模型应用于OLI时序影像进行盐分反演制图,通过叠置方法进行盐渍土演化分析.结果表明,光谱转换方法提高了多传感器间数据一致性.偏最小二乘回归-电导率(PLSR-EC)模型的相关系数为0.700,采用2012年电导率实测值检验该模型反演结果,相关系数为0.690.研究区内高盐分土壤减少并向低盐分土壤转化.Soil salinity quantitative retrieval were investigated based on four Landsat-5 TM(thematic mapper),EO-1(earth observing-one),ALI(advanced land imager),Landsat-8 OLI(operational land imager) time series images and hyper-spectral Hyperion data from 2000 to 2016 for long-term and extensive soil salinity monitoring over the Yellow River Delta.The Hyperion data were resampled to simulate TM,ALI,OLI data according to the corresponding spectral response functions,respectively.The spectral band inter-calibration coefficients were calculated from homologous TM,ALI and OLI bands by the statistical regression method.Then,the TM,ALI images were transformed into the OLI time series images.The PLSR(partial least square regression) model and the MLR(multiple linear regression) model were used to quantify the relationships between the soil spectra and the soil salinity parameters.The optimal predictive model was applied to OLI time series images to map soil salinity.The temporal variations in soil salinity were detected by overlay analysis.The results show that the band inter-calibration method improves the consistency of the multi-senor data.The PLSR-EC(electrical conductivity) model exhibits a correlation of 0.700 and the model validation yields a correlation of 0.690 with the soil samples collected in 2012.High salinity soil in the range of research area reduces and converts to low salinity soil.

关 键 词：盐渍土 黄河三角洲 光谱转换 偏最小二乘回归 动态分析 

分 类 号：TP237[自动化与计算机技术—检测技术与自动化装置]