Spatio-temporal evolution and future scenario prediction of karst rocky desertification based on CA–Markov model

• Published in: Arabian journal of geosciences 2021-07, Vol.14 (13), Article 1262
• Main Authors: Chen, Fei, Wang, Shijie, Bai, Xiaoyong, Liu, Fang, Tian, Yichao, Luo, Guangjie, Li, Qin, Wang, Jingfeng, Wu, Luhua, Cao, Yue, Li, Huiwen, Deng, Yuanhong, Li, Chaojun, Yang, Yujie, Tian, Shiqi, Lu, Qian, Hu, Zheyin, Xi, Huipeng
• Format: Article
• Language: English
• Subjects: Cellular automata; Desertification; Disasters; Distribution; Earth and Environmental Science; Earth science; Earth Sciences; Ecology; Environmental restoration; Evolution; Governance; Ground cover; Karst; Lithology; Markov chains; Original Paper; Population density; Restoration; Simulation; Soil erosion; Spatial distribution; Temporal variations
• ISSN: 1866-7511; 1866-7538
• DOI: 10.1007/s12517-021-07584-4

Although the cellular automata (CA) model has been extensively applied in the simulation of ground cover changes, but it is rarely applied in the simulation of the driving forces of karst rock desertification (KRD). KRD has become one of the most serious ecological disasters in southwest China. Thus, it is necessary to accurately identify the driving factors affecting the occurrence and development of KRD. Accurately predicting the future development trend of KRD has great significance for quantitative evaluation of ecological environment governance and restoration in karst areas. We used the actual interpretation of KRD data in 2011 and 2016, based on the geographical detector to select the driving factors for the occurrence and development of KRD, and used the CA model to simulate the spatial and temporal changes of KRD. Results show that (1) the kappa verification accuracy for all types of KRD was above 0.5 when the CA model was used for the simulation of the spatial distribution of KRD and thus the theoretical requirements for accurate identification of the distribution of KRD were met. (2) Driving factors can be accurately screened by using the geodetector model to analyze the driving factors of KRD. The strengths of the factors follow the order lithology (0.35) > population density (0.30) > slope (0.21) > soil erosion (0.16) > altitude (0.05). (3)The combination of geodetector and the CA–Markov model results in the accurate prediction of the evolution of KRD and reduction in the arbitrariness of artificial subjective selection factors and the possibility of misjudgement. (4) From 2011 to 2021, the total area of KRD in the study area decreased at a rate of 29.96 km 2 ·a −1 , and KRD land indicated an overall trend of improvement. (5) Under the trend of overall improvement of KRD, some areas remain in which KRD increased and worsened. In the process of governance and protection, the impact of such deterioration on ecological environment must be considered.