海岸沙丘沙和海滩沙的粒度识别与区分是海岸风沙研究中争论较大的问题之一。以我国华南海岸为研究区域,基于在海岸风沙地貌广泛发育的闽南、粤东、粤西、海南岛沿岸4个区域的141个现代海岸沙丘沙和70个海滩沙样品的粒度数据,尝试运用多元线性判别分析和人工神经网络分析两种方法进行现代沙丘沙和海滩沙粒度的定量识别。结果表明,比较而言非线性的人工神经网络分析法较传统线性判别分析方法的识别率为高,人工神经网络分析可以通过粒度数据区分部分海岸的沙丘沙和海滩沙,但整体而言2种定量识别方法均难以完全对华南海岸的现代沙丘沙和海滩沙进行粒度的有效区分。
The discrimination of grain size of modern aeolian sands and beach sands is one of the heated argument problems in coastal aeolian research. Based on the data of grain size of 141 modern Aeolian sand samples and 70 beach sand samples from the southern Fujian coast,eastern Guangdong coast,western Guangdong coast and Hainan island coast which are the typical coastal dune distribution places in south China coast,the modern aeolian sands and beach sands are quantitatively discriminated by use of the multivariate linear discriminant analysis method and B-P artificial neural network analysis method. The results show that the B-P artificial neural network analysis method has higher discrimination precision than the multivariate linear discriminant analysis method. Modern aeolian sands and beach sands in Hainan island coast could be discriminated based on the grain size data by use of the B-P artificial neural network analysis method. However,viewing the results as a whole,the modern aeolian sands and beach sands in south China coast could not be significantly discriminated based on grain size by using two quantitative discriminated methods.