Soil moisture and salinity are two crucial coastal saline soil variables, which influence the soil quality and agricultural productivity in the reclaimed coastal region. Accurately characterizing the spatial variability of these soil parameters is critical for the rational development and utilization of tideland resources. In the present study, the spatial variability of soil moisture and salinity in the reclaimed area of Hangzhou gulf, Shangyu City, Zhejiang Province, China, was detected using the data acquired from radar image and the proximal sensor EM38. Soil moisture closely correlates radar scattering coefficient, and a simplified inversion model was built based on a backscattering coefficient extracted from multi-polarization data of ALOS/PALSAR and in situ soil moisture measured by a time domain reflectometer to detect soil moisture variations. The result indicated a higher accuracy of soil moisture inversion by the HH polarization mode than those by the HV mode. Soil salinity is reflected by soil apparent electrical conductivity （ECa）. Further, ECa can be rapidly detected by EM38 equipment in situ linked with GPS for characterizing the spatial variability of soil salinity. Based on the strong spatial variability and interactions of soil moisture and salinity, a cokriging interpolation method with auxiliary variable of backscattering coefficient was adopted to map the spatial variability of ECa. When compared with a map of ECa interpolated by the ordinary kriging method, detail was revealed and the accuracy was increased by 15.3%. The results conclude that the integrating active remote sensing and proximal sensors EM38 are effective and acceptable approaches for rapidly and accurately detecting soil moisture and salinity variability in coastal areas, especially in the subtropical coastal zones of China with frequent heavy cloud cover.
全氮是土壤肥力的重要指标,对作物产量具有决定性作用,采用土壤可见-近红外(Vis-NIR)光谱预测技术及时获取土壤全氮含量信息具有重要意义。采用来自5省的450个土壤样本来验证局部加权回归方法(LWR)结合Vis-NIR光谱技术预测大面积土壤全氮含量的适用性。结果表明,LWR模型的预测效果优于偏最小二乘回归(PLSR)、人工神经网络(ANN)和支持向量机(SVM),选取主成分数为5,相似样本为40时,模型验证的决定系数(RP2)为0.83,均方根误差(RMSEP)为0.25 g kg-1,测定值标准偏差与标准预测误差的比值(RPD)达到2.41。LWR从建模集中选取与验证样本相似的土样作为局部建模样本,降低了差别大的样本对模型的干扰,从而提高了模型的预测能力。因此,LWR建模方法通过大范围、大样本土壤光谱数据进行大尺度区域的全氮等土壤属性预测时能够发挥更好的作用。
