三峡库区柑桔园紫色土光谱特征及其与氮素相关性研究

应用可见光-近红外光谱技术研究了三峡库区柑桔园紫色土的光谱特征及其与土壤氮素营养含量的相关性。结果表明，三峡库区柑桔园土壤光谱反射率在可见光区域随波长增加呈直线上升，在近红外短波区域(780～1 750 nm)基本趋于平稳，波动较小，而在近红外长波区域(1 750～2 400 nm)上波动性和反射率均较大，并且在近红外长波段区域的1 416，1 913，2 209 nm处出现了强的吸收峰。土壤有效氮、全氮含量均与光谱反射率呈正相关，与倒数对数光谱成负相关。在可见光541 nm处，土壤有效氮含量与反射光谱一阶导数微分值达最大正相关，相关系数为0.605**,二者响应的最佳拟合方程为y=2E+09x2-3E+06x+890.49(R2=0.5，x为反射光谱一阶导数值)。在近红外长波段1 909 nm处，土壤全氮含量与反射光谱的倒数对数值的相关性最好，相关系数为-0.612**,二者响应的最佳拟合模型为y=1.372 12-2.107 5x+0.859 2(R2=0.4，x为反射光谱倒数对数值)。

Study on the Correlation of Spectral Characteristic and Nitrogen Content of the Soil in Citrus Orchard of Three Gorges Reservoir Area

The relationship between the spectrum characteristics and nitrogen content of soils in citrus orchard of the Three Gorges Reservoir Area was studied by analyzing the visible-near infrared spectrum. The results showed that the soil reflectivity increased lineally as the wavelength increases across the visible spectrum and reached a stable plateau in the short wavelength near-infrared region（780-1 750 nm）without much fluctuation. In the long wavelength near-infrared region （1 750-2 400 nm）the reflectivity of the soils was higher with higher fluctuation. There were three strong absorbance peaks around 1 416, 1 913 and 2 209 nm, respectively, in the long wavelength infrared region. Soil available nitrogen content and total nitrogen content were positively correlated with soil light reflectivity but negatively correlated with catoptric-spectrum values reciprocal logarithm. At 541 nm of visible light region, a high positive correlation was found between the available nitrogen content and the first derivative of the soil reflective spectrum with a correlation coefficient of +0.605** and the best fitting equation was y=2E+09x2-3E+06x+890.49, where R2=0.5， and x is the first derivative of the soil reflective spectrum. At 1 909 nm of the near-infrared long wavelength region, the correlation between the total nitrogen content and the reciprocal-log values of the reflective spectrum of the soils was the best with a correlation coefficient of -0.612**, and the best fitting equation was y=1.372 1x2-2.107 5x+0.859 2, where R2=0.4, and x is the reciprocal values of the log reflective spectrum of the soils.