Abstract:Saline-alkali land is an important reserve land resource in China. Real-time and accurate acquisition of soil information is important for the classification and evaluation of soil salinization to prevent its degradation and realize agriculture sustainable development. Selecting different soil salt crusts in northern Ningxia Yinchuan as the study objects, based on soil science and geostatistics methods, and taking the spectra data of different soil salt crusts and measured soil salinization parameters in 0~5cm layer of laboratory as the source of information, the characteristics of spectra reflectance of different salt crusts were analyzed, the sensitive spectral wavelengths or index to pH value, EC and salt ions in crust layer were selected, and then the soil salinization monitoring models were established and confirmed. Results showed that the spectral reflectance of white alkali crust was the highest among different soil saline crusts;the reflectance of equine caustic crust was next, and the reflectance of black alkali crust was the lowest. The main salt crust types in the study region could be classified by the spectral reflectance of the field. The highest correlation coefficients between the transformations of smoothing reflectance through the first order differential, the first derivate differential of logarithmic reciprocal of reflectance, continuum removal, the first derivative of continuum removal and salinity parameters were significantly improved than the transformation of smoothing reflectance gradually. The best transformation method of reflectance about soil pH value, EC, CO2-3 and Mg2+ were the first derivative of continuum removal;the best transformation method of reflectance about soil SO2-4, Ca2+, K+ were first derivate differential of logarithmic reciprocal of reflectance;the best transformation method of reflectance about soil HCO-3, Cl- nd Na+ were the first order differential. There was the strongest correlation between the first derivative of continuum removal and salinity parameters. On the whole from different salinity parameters, the sensitive wavelength was 450nm, 470nm and 485nm in blue region;501nm and 575nm in green region;680nm in red region;there were many sensitive wavelengths in infrared region. The highest correlation coefficients between pH value, EC, CO2-3, HCO-3, Cl-, SO2-4, Ca2+, Mg2+, K+ and Na+ and nine salinity indexes was S1 (Salinity index), S3 (Salinity index), SCI (Soil curst index), S3, SI3 (Salinity index 3), S2 (Salinity index), SCI, S2 (Salinity index), SI3 and S3, respectively. Except for CO2-3, the models were suitable for predicting the content of soil pH value, EC, and other salinity parameters, and there was the highest R2 about Na+ n this region. The study would provide some beneficial references for regional soil salinity classification and prediction.
