Integrating remote sensing with GIS-based multi-criteria evaluation approach for Karst rocky desertification assessment in Southwest of China

8th International Symposium of the Digital Earth

地球科学;计算机科学

Zhang, Z.^1 ; Xu, W.^1 ; Zhou, W.^1 ; Zhang, L.^1 ; Xiao, Y.^1 ; Ou, X.^2 ; Ouyang, Z.^1

State Key Laboratory of Urban and Regional Ecology, Research Center for Eco-Environmental Sciences, Chinese Academy of Sciences, Beijing 100085, China^1

Institute of Ecology and Geobotany, Yunnan University, Kunming 650091, China^2

关键词: Clustering methods;    Fuzzy set membership;    Land cover informations;    Multi-criteria evaluation;    Pair-wise comparison;    Qualitative information;    Remote sensing data;    Vulnerable environments;   

Others  :  https://iopscience.iop.org/article/10.1088/1755-1315/18/1/012038/pdf
DOI  :  10.1088/1755-1315/18/1/012038

学科分类：计算机科学（综合）

来源: IOP

【 摘 要 】

The increasing exploitation of Karst resources is leading to severe environmental impacts, as Karst frequently occurs in the most fragile and vulnerable environments. This paper presents a multi-criteria evaluation (MCE) approach in a spatial context to support Karst rocky desertification (KRD) assessment by integrating remote sensing data with GIS. The study area is located in Wenshan Prefecture, Yunnan Province, Southwest of China. Criteria and impact factors for KRD first were identified and weighted through pairwise comparison method. A GIS fuzzy set membership function was then used to generate gradient effects of each criterion, and a clustering method based on K-mean algorithms was used to classify KRD into several descending rank zones (or levels). Both ROC and error matrix assessments indicated that the MCE approach is better than the NDVI approach. In addition, we found it is useful to integrate the topographic and human disturbance factors into KRD mapping and assessment, compared with most of the previous KRD assessment studies mainly focused on developing vegetation or land cover information in karst regions by using remote sensing alone. Furthermore, the integrated MCE approach is robust, flexible, and easy to be implemented. It also explicitly includes the quantitative and qualitative information, for instance, opinions of decision makers and experts as well as characteristics of the landscape.