摘要
在融雪期通过实地雪深与光谱测量,研究了实测光谱反射率与积雪深度之间的关系。分别采用反射率Reflectance、倒数之对数log(1/R)、标准化比值(R/R_(450-750))和Box-Cox转换值4种光谱指标与不同波段组合建立对雪深的多元线性回归预测模型,并利用验证样本集对模型进行了检验。结果表明:野外实测反射光谱与雪深值呈良好的正相关关系,采用可见光波段与近红外波段结合的波段组合预测积雪深度的建模与检验精度均高于单独使用可见光波段或近红外波段,标准化比值结合可见光与近红外波段组合建模精度与检验精度均为最高,4种光谱指标中,倒数之对数和Box-Cox转换值的模型判定系数与检验精度均不理想。
Based on the monitored data of snow depth and measured VIS-NIR reflectance on given spots,the relationship between measured reflectance and snow depth is analyzed.Besides original field-measured spectrum（R）,another spectrum index is also calculated：inverse-log spectrum[lg（1/R）],multivariate linear regression models are built to evaluate snow depth based on the two spectral indices and the model accuracy of snow depth fitting are discussed with validated sample groups.The results show that there is a significant positive correlation between snow depth and original reflectance.Accuracy of the model based on R/R_（450-750） combining VIS-NIR gets the best effect and inverse-log spectrum[lg（1/R）]calculating and box-cox is of less help to improve the predicting effect.
出处
《中国农村水利水电》
北大核心
2016年第5期105-108,共4页
China Rural Water and Hydropower
基金
国家自然科学基金面上项目(41171023)
关键词
光谱预处理形式
融雪期
雪深
多元线性回归预测模型
spectral data pretreatment
melt period
snow depth
multiple linear regression forecasting model
