Cell-level coupling of a mechanistic model to cellular automata for improving land simulation. Issue 1 (31st December 2023)

• Record Type: Journal Article
• Authors: Chen, Guangzhao; Zhuang, Haoming; Liu, Xiaoping
• Is Part Of: GIScience & remote sensing. Volume 60:Issue 1(2023)
• Publication Date: 2023-12-31
• Subjects: Cellular automata -- mechanistic model -- coupled model -- dynamic vegetation model -- FLUS model -- land use and land cover change simulation; Geodesy -- Periodicals; Cartography -- Periodicals; Aerial photogrammetry -- Periodicals; Remote sensing -- Periodicals; 526.05
• DOI: 10.1080/15481603.2023.2166443
• Languages: English
• ISSNs: 1548-1603

Abstract

ABSTRACT: The land use and land cover change (LUCC) process is crucial for climate and environmental change studies. Cellular automata (CA) models based on machine learning techniques have commonly been used to simulate LUCC. However, conventional CA models can only capture the historical relationship between driving forces and LUCC through statistics. Climate change significantly affects LUCC, especially natural vegetation. The statistical relationship between driving forces and LUCC may change with the climate. Furthermore, the existing coupled models of CA and mechanistic models are only loosely coupled in terms of the quantity of land demand provided by mechanistic models, which denotes that mechanistic models do not directly guide CA models in simulating spatial land dynamics. Thus, herein, a novel model that couples Lund Potsdam Jena (LPJ), a mechanistic model, and CA at the cell level is proposed to consider the mechanisms and statistics in LUCC simulations. The proposed coupled model (LPJ-FLUS) is validated by comparison with two historical LUCC simulations in China from 2001 to 2010 and 2015. The results show that the proposed coupled model affords higher simulation accuracy, especially on natural vegetation, compared to the two conventional CA models. The overall Figure of Merit value of the LPJ-FLUS model in the historical LUCC simulations is about 5% higher than that of the conventional CA model, reaching 23.37%, and is even 8% higher for some natural vegetation types. The proposed model is employed to predict the future LUCC under two SSP-RCP scenarios in China from 2015 to 2100. The simulation results show that the proposed model effectively combines the strengths of LPJ and CA, retaining the high spatial resolution of CA while representing the spatial possibilities of LUCC under different future climate scenarios assessed by LPJ. The quantitative validation results show that the simulation results of the LPJ-FLUS model are more spatially correlated with the LPJ model than those of conventional CA models. The proposed LPJ-FLUS model is a valuable attempt toward tightly coupling mechanistic and CA models at the cell level. Additionally, it has promising potential for high spatial resolution LUCC simulations and environmental impact analysis under future extreme climate scenarios.