论文部分内容阅读
以不同含水量的宁夏典型龟裂碱土为研究对象,系统分析了土壤光谱与土壤含水量的相关性,并建立了含水量预测模型.结果表明:随着含水量的增加,土壤光谱反射率逐渐降低,当土壤含水量高于田间持水量时,土壤光谱反射率随着含水量的增加呈增加趋势.土壤光谱反射率原始数据(r)、平滑后的反射率(R)和反射率对数(lgR)与龟裂碱土水分含量呈极显著负相关关系,整个波段R与土壤水分含量的相关系数平均比r和lgR分别高0.0013和0.0397;反射率倒数(1/R)和反射率倒数的对数[lg(1/R)]2种变换形式与龟裂碱土水分含量呈正相关关系,在950~1000 nm的相关系数平均比400~950 nm高0.2350;3种一阶微分变换形式与土壤水分的相关性不稳定.基于r、lg(1/R)、反射率的一阶微分R’和反射率对数的一阶微分(lgR)’采用不同回归模式建立的龟裂碱土含水量预测模型平均决定系数分别为0.7610、0.8184、0.8524和0.8255,其中R’的幂函数模式决定系数高达0.9447,该模型预测的土壤含水量与室内实测值拟合度为0.8279,说明该模型预测精度最高,采用r建立的模型预测精度最低.研究结果可为龟裂碱土含水量预测和当地农田灌溉提供科学依据.
Taking typical Ningxia calcareous alkaline earth with different water content as the research object, the correlation between soil spectrum and soil water content was systematically analyzed and a water content prediction model was established. The results showed that the spectral reflectance gradually increased with the increase of water content , The soil spectral reflectance increased with the increase of water content when the soil water content was higher than the field water holding capacity.The original data of soil spectral reflectance (r), smoothed reflectance (R) and logarithm of reflectance (lgR) had a very significant negative correlation with the water content of the alkaline soil, the average correlation coefficient of R and soil water content was 0.0013 and 0.0397, respectively; the reciprocal of reflectance (1 / R) and the reciprocal of reflectance Logarithm [lg (1 / R)] showed a positive correlation with the water content of alkaline earth, and the correlation coefficients at 950-1000 nm averaged 0.2350 higher than those at 400-950 nm. Three kinds of first-order differential transformations and soil The correlation of water content is not stable. Based on r, lg (1 / R), the first order differential R ’of reflectivity and the first order differential of logarithm of reflectivity (lgR)’ were predicted by different regression models The average coefficient of determination of the model is 0.7610,0.8184,0.8 respectively 524 and 0.8255, respectively, of which the power function mode of R ’determines the coefficient as high as 0.9447. The predicted soil moisture content of the model is 0.8279, which shows that the prediction accuracy of this model is the highest and that of the model established by r is the lowest. The results provide a scientific basis for predicting the water content of the alkaline soil and the irrigation of local farmland.