叶面降尘对香梨叶片高光谱特征的影响及定量反演研究

叶面降尘指大气中的浮尘经重力沉降后，在植物叶片表面所形成的一层明显积尘，对其进行监测，可为沙尘区的环境评价及农业灾害评估提供基本依据。在量化叶面降尘的基础上，研究了叶面降尘对南疆香梨叶片高光谱特征的影响，分析了叶面降尘与反射率的相关性，并建立了叶面降尘的高光谱定量反演模型。研究结果表明，叶面降尘使可见光(400~700 nm)反射率增加，最大变幅位于666 nm，绝对变化率为-10.50%，相对变化率为62.89%;使近红外(701~1 050 nm)的反射率降低，最大变幅位于758 nm，绝对变化率为12.04%，相对变化率为-41.75%。叶面降尘量大于20 g·m-2时，叶片除尘后，绿峰、红光吸收谷、蓝光吸收谷得到凸现，500~750 nm波段的斜率明显变大。叶面降尘量低于20 g·m-2时，其对绿峰的形状和面积影响不大。叶面降尘与反射率在可见光波段呈正相关，与近红外波段呈负相关，可见光波段的相关性要优于近红外波段，最大相关系数(0.61)出现在663 nm。在构建的七种PLSR反演模型中，倒数对数一阶微分模型具有较好的稳定性及预测能力，决定系数(R2)、均方根误差(RMSE)、预测方差比(RPD)分别为0.78，3.37和2.09，对叶面降尘具有很好的预测能力，其余模型的RPD均小于2.0。研究结果为叶面降尘的高光谱遥感监测提供了一定的理论依据，同时为沙尘区环境评价及农业灾害评估提供了新的数据获取方法与思路。

Effect of Foliar Dustfall Content (FDC) on High Spectral Characteristics of Pear Leaves and Remote Sensing Quantitative Inversion of FDC

The precipitation of floating and sinking dust on leaves of plants is called as foliar dustfall. To monitor foliar dustfallIt, it will provide fundamental basis for environmental assessment and agricultural disaster evaluation of dust area. Therefore, the aim of this work to (1) study the effect of foliar dustfall content (FDC) on high spectral characteristics of pear leaves, (2) analyze the relationship between reflectances and FDC, and (3) establish high spectral remote sensing quantitative inversion model of FDC. The results showed that FDC increased reflectances of visible band (400~700 nm) with maximum band of 666 nm. Absolute and relative rates of change were -10.50% and -62.89%, respectively. The FDC decreased reflectances of near infrared band (701~1 050 nm) with maximum band of 758 nm. Absolute and relative rates of change were 12.04% and 41.75%, respectively. After dustfall was removed, reflection peak of green light and absorption valley of red and blue light became prominent, and slope of 500~750 nm wave band increased when FDC was more than 20 g·m-2. While FDC just slightly affected shape and area of reflection peak of green light when FDC was less than 20 g·m-2. FDC were positive and negative correlated with reflectances of visible band and near infrared band, respectively. Maximum correlation coefficient (0.61) showed at 663 nm. All of 7 inversion models, the model based on the first-order differential of logarithm of the reciprocal had better stability and predictive ability. The coefficient of determination(R2), root mean square error (RMSE) and relative percent deviation (RPD) of this model were 0.78, 3.37 and 2.09, respectively. The results of this study can provide a certain reference basis for hyperspectral remote sensing of FDC.