基于光谱水分指数的阔叶树种叶片等效水厚度估算

Leaf equivalent water thickness estimation based on spectral moisture indexes in broadleaf species

植物叶片的水分含量是评价植物水分状态的一项重要指标。为快速、准确地获取植物的叶片含水量,分别使用传统敏感性分析方法和扩展傅立叶振幅灵敏度分析(EFAST方法)对PROSPECT模型进行了敏感性分析。在此基础上筛选与叶片含水量敏感度高的光谱水分指数,并以阔叶树种为研究对象,根据地面实测叶片等效水厚度(lEWT)和光谱数据建立了叶片等效水厚度估算模型并验证。结果表明:EFAST方法比传统的敏感性分析方法,可以量化叶片各项生化参数的敏感特征。叶片等效水厚度的敏感波段为900~2 500 nm,但在这个范围内光谱反射率也受叶片结构(N)和干物质含量(Cm)的影响;在选取的8个光谱水分指数中,水分胁迫指数(IMSI)、归一化差异红外指数(INDII)和归一化差异水分指数(INDWI1640)可以有效提高lEWT的敏感度。基于实测数据对植被水分指数建立的叶片等效水厚度估算模型的验证结果显示:INDWI1640和INDII估算模型的预测精度要优于其他指数。对比分析表明:在选取的光谱水分指数中,INDWI1640和INDII是叶片尺度上估算水分的最佳指数,而归一化差异水分指数(INDWI1240)和简单水分指数(ISRWI)适合于低值区域的叶片等效水厚度反演。

To quickly and accurately obtain the leaf water content of plants,an important evaluation indicator for evaluating the plant water status,a sensitivity analysis with the PROSPECT Model was performed using the traditional sensitivity analysis method and the EFAST(Extended Fourier Amplitude Sensitivity Test)method.Then,the spectral water indexes with a high sensitivity were selected.Using 8 deciduous broadleaf tree species as the research samples,1 494 spectral data as input data,an equivalent water thickness estimation model was established and validated based on the ground equivalent water thickness and spectral data.Results showed that EFAST was more sensitive than the traditional sensitivity analysis method in quantifying characteristics of various biochemical parameters.The sensitive wavelength of leaf equivalent water thickness(lEWT)was in the range of 900-2 500 nm,but in the same range,spectral reflectance was also affected by leaf structure and dry matter content.Among the eight spectral water indexes selected,moisture stress index(IMSI),normalized difference infrared index(INDII),and normalized difference water index(INDWI1640)effectively increased sensitivity to lEWT.Validation results based on measured data showed that prediction accuracy of INDWI1640 and INDII estimation models were better than other indexes.Therefore,among the spectral water indexes selected for this study,INDWI1640 and INDII were the best indexes for estimating leaf equivalent water thickness at the scale of the plant leaf;whereas,normalized difference water index(INDWI1240)and simple ratio water index(ISRWI)were suitable for inversion of leaf equivalent water thickness in the areas with lower moisture.The research demonstrated that EFAST method has significant value in guiding modification for localization of model parameters.