基于实测光谱的叶绿素a估测模型

Chlorophyll-a concentration estimation model based on field spectral reflectance of water

为了有效地监测太湖湖区的富营养化程度，建立一种最佳的估测模型。对太湖水体进行了连续监测，获得了从2004年10月20～29日的实测光谱数据和水质数据。通过实测光谱反射率和叶绿素a浓度的相关分析，得到了叶绿素a反演的最佳波段，并通过比值法、反射峰位置法、峰谷距离法、峰高法和一阶微分法建立了基于实测光谱的叶绿素a浓度的估测模型。通过14个检验点，分别对5种模型进行了对比分析，结果发现，通过比值法建立的估测模型总体误差最小，估测精度最高，其695nm附近反射峰与677nm反射谷的比值与叶绿素a浓度之间具有良好的相关性，可决系数R^2可达0．7433，可以作为整个湖区叶绿素a浓度的最佳估测模型。最后，应用该模型对该季节湖区的营养化程度做出了总体评价。

In order to monitor trophic level of Tai Lake effectively and build a best estimation model, the spectral reflectance data and water quality data were acquired on 20 -29 October 2004 after continuous monito- ring. According to the correlation analysis between the field spectral reflectance and chlorophyll-a, a best inversion band was found, and the chlorophyll-a concentration estimation model was built based on field spectral reflectance by the method of ration, the position of reflectance peak, distance of peak and valley, peak and the first order differential. Based on the practical data collected from 14 positions in Tai Lake, five models were set up. The numerical calculations of the models show that the model built by the ratio method has the lowest error, the ratio of reflectance peak near the 695 nm and reflectance valley near the 677 nm has good correlation with chlorophyll-a concentration and the coefficient of determination is 0. 7433. Therefore, the model can be as the best model to estimate chlorophyll-a concentration. Finally, this model was used to evaluate the trophic level of Lake in this season.