基于正负地形的喀斯特地貌分类研究——以贵州喀斯特区为例

Research on karst landform classification based on positive and negative terrains——Taking the karst region of Guizhou Province as an example

地形因子具有直观性、可视性,常被作为喀斯特地貌分类的主要依据。已有关于喀斯特地貌分类研究所用的地形因子较多且复杂,分类特征不易被公众学习和理解,导致成果的应用和推广难度较大。因此以贵州喀斯特区为研究对象,运用地形开度法和随机森林模型,筛选重要性正负地形指标,探讨喀斯特地貌分类方法。结果表明:1)贵州喀斯特区的蚕食度均值为1.35,深切度均值为215.36 m,平均粗糙度比均值为1.01,负地形面积大于正地形面积,正负地形不规则且破碎度较大,正地形被沟谷深切明显,地貌的异质性强烈。2)蚕食度和深切度2个指标的重要性指数分别为0.93和1.16,能较好地反映喀斯特地貌的差异性。3)根据蚕食度和深切度贵州喀斯特地貌分为23类,分类精度优良达93.33%。在不同的喀斯特地貌区,因为蚕食度和深切度不同,导致水、土、光、热资源丰度不同,因而它们的土地利用方式各异。研究简化了分类判别喀斯特地貌的地形指标,降低了公众对喀斯特地貌的认识和理解难度,并为数字地形技术在喀斯特地貌分类中的推广应用提供参考。

Topographic factors are intuitive and visual,and are often used as the main basis for karst landform classification.The topographic factors used in existing studies on karst landform classification are numerous and complex,and the classification features are not easy for public's learning and understanding,which makes the application and promotion of the results difficult.The study takes the karst region of Guizhou Province as the study area,selects the important positive and negative terrain indicators and explores the classification method of karst landform based on the terrain openness method and random forest algorithm.The results show that:1)In the karst region of Guizhou Province,the average nibbling degree is 1.35,while the average cutting depth is 215.36 m and the average mean-roughness-ratio is 1.01,the area of negative terrain is larger than the area of positive terrain,the positive and negative terrains are irregular and fragmented,and the positive terrain is cut by gully and valley obviously,and the heterogeneity of landform is strong.2)The importance indices of the two indicators,nibbling degree and cutting depth,are 0.93 and 1.16 respectively,which can better reflect the heterogeneity of karst landform.3)According to the two indicators of nibbling degrec and cutting depth,the landform of karst region of Guizhou Province can be classified into 23 categories,with an accuracy of 93.33%.In different karst landforms,due to different levels of nibble degree and cutting depth,the abundance of soil,water,light and heat resources varies,resulting in different land use methods.Through this study,the terrain indicators used to classify and discriminate karst landforms are simplified,which reduces the difficulty of the public's learning and understanding on karst landforms,and provides a reference for the popularisation and application of the digital terrain technology in the classification of karst landforms.