Simulating and assessing the evolution of collapsing gullies based on cellular automata-Markov and landscape pattern metrics: a case study in Southern China

Abstract

Collapsing gullies, which involve considerable erosion and extreme landform changes, frequently occur in the granite region of Southern China. Capturing the evolution of collapsing gullies is useful and effective for predicting erosion amount and landform changes. However, the evolution of collapsing gullies is too complex to simulate using conventional models. The aim of this study is to modify the traditional cellular automata (CA)-Markov model for simulating the evolution of collapsing gullies and then quantify their morphology using landscape pattern metrics. A hillslope eroded by collapsing gullies located in Longmen Town of the granite region of Southern China is used as a case study. Three digital elevation models (DEMs) were derived on 11 March 2017, 21 July 2017, and 2 December 2017 from a remotely piloted vehicle and global positioning system (GPS) real-time kinematics. Rainfall data for the corresponding time was recorded by a tipping-bucket rain gauge. Using these data, the CA-Markov model for simulating the evolution of collapsing gullies was developed, and then the most accurate one was chosen to predict the evolution on 2 December 2018. Evolution of the case study hillslope was interpreted and assessed using landscape metrics to capture the erosional trends of collapsing gullies. The area differences of the modified CA-Markov model are lower than those of the traditional model while the kappa coefficients of the modified CA-Markov model are higher than those of the traditional model; that is, the modified CA-Markov model performs better for simulating and predicting the evolution of collapsing gullies. The kappa coefficients also demonstrate that both scouring and gravity impact collapsing gullies, and scouring force is more effective than gravity. Based on the evolution prediction, the erosion amount of collapsing gullies in the subsequent year is predicted to be 904.1 m3. Spatial pattern analysis showed that the mid-lower part of a hillslope eroded by collapsing gullies will continue to be intensively eroded and the ground surface will become more fragmented. The use of a modified CA-Markov model and landscape pattern metrics provide an improved and effective method for understanding the spatial and temporal variations of collapsing gullies landform and ground surface, as well as better capturing the erosional trends of collapsing gullies.