基于遥感数据融合的黄河三角洲土壤盐分时空变化研究

  目的  研究2005 ~ 2018年黄河三角洲地区土壤盐分在年内和年际尺度上的时空变化特征。  方法  基于2005 ~ 2018年春季覆盖黄河三角洲地区的MODIS和Landsat系列数据，采用增强型自适应反射率时空融合模型（ESTARFM）获得30米分辨率高频地表反射率数据。基于2005年实测土壤盐分数据和Landsat地表反射率数据，采用随机森林方法建立土壤盐分反演模型，反演2005 ~ 2018年黄河三角洲地区春季土壤盐分数据，分析土壤盐分含量的时空演变特征。  结果  ESTARFM融合数据具有较为理想的精度，地表反射率总体误差在4%以内。年内尺度上，2 ~ 4月份黄河三角洲地区土壤盐分含量呈总体下降趋势，3 ~ 4月份存在盐分含量短期回升现象，进入4月份后，土壤盐分含量明显下降，非盐渍土和轻度盐渍土占比增加。年际尺度上，2005 ~ 2018年研究区土壤盐分含量呈先升后降趋势，最大值出现在2009年（4.262 g kg?1），最小值出现在2005年（3.604 g kg?1）。2009年以来，研究区内非盐渍土和轻度盐渍土面积显著增加，盐土面积显著减少，盐渍化程度明显改善。  结论  增强型自适应反射率时空融合模型可用于高频次土壤盐分数据反演，反演结果可加深对土壤盐分年内和年际变化规律的认识。

Investigation of Spatio-temporal Variations of Soil Salinization in the Yellow River Delta Based on Remote Sensing Data Fusion Technique

  Objective  The spatio-temporal variations of soil salt content (SSC) at seasonal and inter-annual scales will be inversed in the Yellow River Delta from 2005 to 2018.   Method  This study derived 30-m resolution high-frequency surface reflectance data over the Yellow River Delta from 2005 to 2018 with the integration of the Moderate-resolution Imaging Spectroradiometer (MODIS) and Landsat series sensors data via the Enhanced Spatial and Temporal Adaptive Reflectance Fusion Model (ESTARFM). Based on soil sampling data in 2005 and the Landsat-5 Thematic Mapper (TM) surface reflectance data, a random forest model was established to model the relationship between SSC and spectral reflectance. The model was used to estimate multi-temporal SSC data from 2005 to 2018, based on which the spatio-temporal variations in SSC were analyzed.   Result  The ESTARFM performed well for deriving Landsat-like reflectance data with an overall uncertainty < 4%. On seasonal scales, SSC showed a downward trend from February to April, with an occasional short-term rise in SSC from March to April. From April onwards, SSC decreased significantly, shown as the increasing proportions of non-saline soils and slightly saline soils. On inter-annual scales, SSC first increased and then decreased from 2005–2018. The highest SSC value appeared in 2009 (4.262 g kg?1), and the lowest SSC value appeared in 2005 (3.604 g kg?1). Since 2009, the area of slightly saline soils has increased significantly, which means a substantial improvement in soil salinization.   Conclusion  The ESTARFM method can be used for high-frequency SSC mapping, which promotes our understanding towards intra- and inter-annual dynamics of soil salinity.