土壤碱化严重威胁干旱区农牧业的发展, 传统碱化程度的测定为破坏性取样和实验室分析。 为验证利用普通数码相机的可见光光谱快速准确的估测土壤碱化程度的可行性, 以土壤pH、 野外实测数字照片、 端元光谱数据为基础, 分析了颜色空间(RGB, HLS, CIEL*a*b*)各参数与土壤pH的相关性, 采用偏最小二乘法(PLSR)分别建立了三种颜色空间预测土壤pH的定量模型, 并比较了照片数据与端元光谱数据所建立的模型精度的差异。 结果表明: 三种颜色空间中, 虽然大部分参数与土壤pH间均达到了极显著相关水平, 但CIEL*a*b*空间各参数与土壤pH的相关性最高, 预测模型判定系数R2最高(0795), 并且RMSECV最低(0084)。 预测集也说明了该模型具有较好的精度和稳定性(R2=0781, RMSEP=0158)。 由于CIEL*a*b*颜色空间各参数间数据冗余小, 又避免了RGB空间各参数数值会受设备影响的不足。 因此, 利用其定量提取土壤碱化信息具有一定优势。 数字照片数据与端元光谱数据所建立的预测土壤pH定量模型精度无显著差异。 因此, 数字照片具有提取土壤碱化信息的潜力。

Soil saline-alkalization is one of the most important problems of land degradation and the basic environmental problem in arid and semi-arid regions. The digital photography technology can rapidly and timely provide the information about properties, geographical distribution and extent of soil saline-alkalization. For verifying use digital photography assess degrees of sodality promptly and accurately, based on the monitored data of soil pH and measured VIS-NIR reflectance and photographs on given spots, The correlation were analyzed between soil pH and color space model parameters, Partial least squares Regression (PISR) was employed to build predicting model of pH value and the different between two Kinds of data were compared. The results showed that most of parameters with significant correlation While the CIEL*a*b* color model was the best. and it is the best model to assess soil pH(R2=0795, RMSECV=0084). Prediction set has also seen it was accurate and stability (R2=0781, RMSEP=0158). The prediction had no significant difference between the digital photography and VIS-NIR reflectance data. The digital image color analysis method showed the potential of being used in soil pH value assessing in the future.