基于小圩形态指数的宣芜平原圩田景观 肌理量化研究——以固城湖永丰圩为例

在当代乡村农耕景观高质量发展与提质增效背景下，以量化分析技术精准研判圩区景观风貌，利用小圩形态指数分析解译典型圩田景观肌理成为乡村景观研究纵深发展的基本问题之一。面向典型圩田肌理特征量化研究需要，综合景观肌理研究的既有成果，确定小圩延伸率、矩形紧凑度、凹凸度、椭圆形状指数和分形维数5项标识性形态指数。以南京固城湖永丰圩为例，将其小圩二维边界转译为指数数据，选取SOM神经网络、K-means等机器学习方法构建圩田肌理识别模型并对样本圩区进行集成解析，实现圩田肌理的表征识别与聚类分析。相关结果为圩田景观肌理研究提供了一种小圩形态指数聚类分析方法，为深入探索乡村景观肌理统计测量技术提供研究参考。

Quantitative Research of Landscape Texture for Polders in Xuanwu Plain Based on Morphological Index: A Case Study of Yongfeng Polder, Gucheng Lake

In the contemporary context of high-quality development and improving?quality and efficiency of farming landscape in rural area, the accurate identification of polder scenery by quantitative analysis technology, and digital interpretation of typical polder landscape texture by polder morphological index has become one of the fundamental issues for profound study of rural landscape. In order to quantify the texture characteristics of typical polder fields, five morphological indices such as polder extension rate, compactness, concavity, elliptical shape and fractal dimension are selected with? the?relevant?research?results, and applied to describe the morphology of single polder. Then, the paper takes Yongfeng polder of Gucheng Lake in Nanjing as an example, the two-dimensional polder boundary data has been collected and translated into index data. Subsequently, machine learning methods including SOM neural network and K-means are selected to construct the polder texture identification model. Finally, the five single polder morphological indices are collaboratively analyzed to achieve the characterization recognition and cluster analysis of polder landscape texture. The paper provides a basic method of cluster analysis using morphological index for polder texture research, and in-depth research suggestions of statistical measurement technology for rural landscape texture.