基于函数拟合的玉米叶片反射光谱时间变化规律分析

为了探索玉米(Zea mays L.)叶片反射光谱时间变化规律，对21个叶位的玉米叶片进行了每天一次的反射光谱测量，获得了玉米活体叶片整个生命周期反射光谱数据1 261条。在此基础上采用光谱相关图方法，对400～960 nm的可见光、近红外波段的玉米叶片反射光谱进行分段拟合，获得了7个表征玉米叶片反射光谱时间变化规律的拟合参数，并对这7个参数的时间变化趋势采用二元二次多项式进行拟合。结果表明该方法对单片叶片反射光谱的拟合效果非常理想，其中98.7%的叶片反射光谱拟合复相关系数r大于0.99，80.9%的叶片反射光谱拟合均方根误差RMSE小于0.001 5。将所有数据的拟合结果与原始数据进行比较，复相关系数r为0.9978，均方根误差RMSE为0.010 5，拟合结果表明该方法较好地保持了反射光谱时间变化的趋势，能够充分利用叶片反射光谱的相关性，有效地表征玉米叶片反射光谱随时间变化的规律。

Temporal Variation Analysis for Spectral Reflectance of Maize Leaves Using a Fitting Method

The present study is to investigate the temporal variation patterns through the dataset of spectral reflectance of maize leaf using a fitting method. In the field experiment, 1 261 pieces of spectral reflectance of maize leaves at different leaf positions were measured every day during its life cycle. After signal/noise analysis, the visible and near infrared (VNIR) band (400～960 nm) was selected in this study. The spectral reflectance was fitted using a spectral scatter diagram (SSD) method. Seven fitting coefficients were employed to denote the temporal variation patterns of maize leaf, which can also be fitted by bi-variate quadratic functions. The comparison of the fitted results with the original measurement data shows that the fitting results are reasonably good, where for 98.7% leaves r is larger than 0.99, and for 80.9% leaves RMSE is less than 0.001 5. All the fitted spectral reflectance is compared with the original measurement data, where r is 0.997 8, and RMSE is 0.0105. The results show that this method is particularly suitable for presenting the temporal variation patterns of the spectral reflectance of maize leaves.