In recent years, soil salinization in the Yellow River Delta under the effects of hydrology, climate and human activities have become increasingly prominent. Based on the 20 Landsat series images of Hekou, Kenli, Dongying districts and Lijin County of Dongying City selected from 1985 to 2018, numerical regression correction method was used to perform image spectral consistency conversion. The partial least squares regression method was used to construct quantitative inversion models of soil salt content. The soil salt content of the study area were retrieved by the best salt prediction model. The temporal and spatial characteristics of soil salt changes in the Yellow River Delta were analyzed. The results showed that the soil salt inversion model constructed with 10 sensitive spectral indices performed higher prediction accuracy, with coefficient of determination R2=0.769 and RMSE=1.125 for calibration, R2=0.752 and RMSE=1.203 for validation, and relative prediction deviation (RPD)=2.08. Using the measured soil salt data in 2016 to verify the inversion accuracy of the model, the correlation between the measured value and the inverted value was 0.7279. The model was used to map the soil salinity of the Yellow River Delta based on 20 images from 1985 to 2018. The abnormal soil salinity retrieval values was all less than 10%. During the study period, the soil salinity showed an overall trend of rising first and then falling which was lowest in 1985 (3.14 g·kg-1) and highest in 1995 (5.86 g·kg-1). Spatially, the area of heavily saline soil and saline soil in the study area decreased, and that of mildly and moderately saline soil significantly increased (66.6%). The total area of saline soil showed an increasing trend. The effects of hydrological and climatic conditions on soil salinity exhibited hysteresis. The increases of temperature promoted soil salinity, with the relationship between the soil salinity and the average temperatures in the past six months and one year being significantly correlated (R=0.507 and 0.538). Soil salinity did not correlate with regional precipitation, and was most affected by the Yellow River streamflow in the previous season (R=-0.543).
近年来,黄河三角洲在水文气候和人类活动影响下的土壤盐渍化问题日益突出。本研究以东营市河口区、垦利区、东营区和利津县为研究区,选取1985—2018年间20期Landsat系列影像,利用数值回归校正法进行影像光谱一致性转换,在此基础上运用偏最小二乘回归法构建土壤盐分定量反演模型;根据最佳盐分预测模型反演的研究区土壤盐分含量数据,分析土壤盐分(SC)时空变化特征,并探讨水文气候因素的影响。结果表明: 选用10个敏感光谱指数构建的土壤盐分反演模型的预测精度较高,其建模决定系数(R2)、均方根误差(RMSE)为0.769、1.125,验证R2、RMSE和相对分析误差(RPD)分别为0.752、1.203、2.08。利用2016年土壤盐分数据进行反演精度检验,实测值与反演值的R2为0.7279。该模型对1985—2018年20期影像进行土壤盐分反演的结果异常值在10%以内。研究期间,区域平均盐分含量总体呈先上升后下降的趋势,最低的年份为1985年(3.14 g·kg-1),最高的年份为1995年(5.86 g·kg-1)。空间变化上,研究区重度盐渍土和盐土面积减少,轻、中度盐渍土面积显著增加(66.6%),盐渍土总面积呈扩大趋势。水文气候条件对土壤盐分的影响具有一定滞后性。气温对土壤盐分积累有促进作用,前半年和前1年的平均气温与含盐量均显著相关,R分别为0.507和0.538;土壤含盐量与区域降水量的相关性不显著,受前季黄河径流量的影响最大,R达-0.543。.
Keywords: Yellow River Delta; climate; hydrology; soil salinity; spatiotemporal variation.