基于高光谱的草地可燃物含水率估测方法

Estimation of Moisture Content of Grassland Fuel Based on Hyperspectra

用美国ASD公司的FieldSpee Pro FR^(TM)光谱仪获得了防火期内含水率不同的草地可燃物的光谱数据,用烘干称重法获得了样品的含水率数据,并用包络线去除法处理样品的光谱反射率曲线。将1400nm附近范围作为水分吸收特征波段,计算出了样品的光谱吸收指数、吸收深度指数、曲线长度和曲线面积指数等特征参数,然后对其与实测可燃物含水率的相关性进行了分析。结果发现,曲线长度指数与含水率的相关系数最高,为0.77。通过用曲线长度指数和含水率进行回归分析,建立了数学模型并对其进行了检验。检验结果表明,该模型的预测相关系数r达到0.80以上。这种草地可燃物含水率快速检测方法是用于预测草原火险的一条快速而又直接的新途径。

A FieldSpee Pro FR^TM spectrograph produced by ASD company in USA is used to obtain the spectral data of the grassland fuel with different moisture content during fire prevention period.A dry weight method is used to get the moisture content of the samples.The continuum removal is used to process the spectral reflectance curves.When using the area near 1400 nm as the wavelength range related to water absorption characteristics,the spectral absorption index,spectral absorption depth index,curve length and curve area etc.of the samples are calculated.Then,the correlation of the calculated results with the measured moisture content of the fuel is analyzed.It is found that the correlation of the curve length index with the moisture content is highest,reaching 0.77.By making a regression analysis of curve length index and moisture content,a mathematic model is established and tested.The test result shows that the model has its predictive correlation coefficient r up to 0.80.This method can detect the moisture content of grassland fuel quickly.It is a new way to predict prairie fire quickly and directly.