淤泥质潮滩沉积物含水量遥感反演模型研究

Remote Sensing Inverse Model for Sediment Water Content in the Muddy Intertidal Flat

潮滩沉积物含水量是表征潮滩地物信息的一个重要特征量,它影响着遥感提取潮滩地物信息的准确性。利用含水量-反射率指数衰减模式建立单波段模型,同时参考4种植被指数的形式构建含水量指数,建立适用于潮滩环境的含水量反演波段组合模型,实现了沉积物含水量的遥感反演。研究结果显示:对反射率数据利用干土反射率进行归一化处理后构建的NEWCI(Normalized Enhanced Water Content Index)归一化增强型含水量指数模型的含水量反演效果最佳。在含水量结果中,除了砂类别反演的确定性系数R2为0.675以外,粉砂、砂质粉砂、粉砂质砂、粘土质粉砂等4种沉积物类型反演的确定性系数均在0.83以上;当不考虑沉积物类型时,NEWCI指数模型的反演精度也达到0.8,RMSE为0.059,表明NEWCI指数模型对沉积物类型不敏感,且能够保持较高的含水量模拟精度,因而适合潮滩沉积物含水量的大范围预测。其反演结果可在潮滩地物信息遥感反演时有效削弱水分的影响,进而提高地物信息提取的效率。

The water content in sediments is one of the important physical parameters for characterizing the surface features of the intertidal flat and affects the accuracy of extracting the surface features of the inter-tidal flat by using remote sensing technology. For reproducing the sediment water content with remote sensing, two models are established： one is a single-band model established based on the exponential atten-uation mode between the water content and the reflectance of the sediments; another is a combinative band model, which is established by constructing the water content index with the reference of the forms of four kinds of vegetation indices. These models are effectively applied to the prediction of water content in the sediments of the muddy intertidal flat. The detailed results show that the model based on the normalized enhanced water content index （NEWCI）, which is built up after normalizing the reflectance data with the reflectance of dried sediments, has the best predictive capacity in the water content inversion. Of the water content inverse results obtained with this model, the deterministic coefficient R2 is about 0. 675 for sands and larger than 0.83 for other four types of sediments, i.e. silt, sandy silt, silty sand and clayey silt. Even without the consideration of sediment types, the inverse accuracy can also reach to 0.8 by using the NEWCI-based model, with the RMSE （Root Mean Square Error） being only 0. 059. These results indicate that the NEWCI-based model is not sensitive to the sediment types. It can keep a high accuracy of the wa- ter content simulation and is suitable for predicting the sediment water content in a large scale in the inter-tidal flat. Its inverse results can effectively reduce the influence of surface water on sediment reflectance, hence improving the efficiency of extracting the surface features by using remote sensing.