Land-use change modeling with cellular automata using land natural evolution unit.

• Proposed a new concept of determining cellular units using the natural evolutionary unit of land (LNU). • LNU can reflect the spatial characteristics of land natural evolutionary process more truly. • Proposed a calculation method for identifying LNU efficiently based on the geographical similarity principle. • The LNU-CA has better simulation effect than the traditional CA in modeling land-use change. Cellular automata (CA) is an important model to simulate land-use change, and the cellular unit can influence the organizational structure, scale selection, and evolutionary ability of CA through its size and shape. Therefore, determining a reasonable cellular unit is essential for land-use change modeling using CA models. However, existing methods used to define the cellular unit are inadequate in their description of the characteristics of the natural evolution of land use, leading to a mismatch between the unit used for simulation and the unit of actual land-use change. To address this issue, in this study, we propose a method for determining the cellular unit based on land natural evolutionary unit (LNU). First, the geographic similarity principle was used as the theoretical basis for this new method. Under this similarity basis, a CA modeling method based on coupling the land natural evolution unit with evolution behavior was then designed to improve the natural evolution of land use. Finally, a case study of the Chenggong District of Kunming City, Yunnan Province, China, was conducted to evaluate the effects of using LNU-defined cellular units on simulating land-use change. The experimental results from the case study show that, simulation of land-use changes using the CA model based on the land natural evolution unit (LNU-CA) exhibits better spatial distribution consistency, higher simulation accuracy with the Kappa coefficient increased by 5.11% and the FOM (Figure of Merit) increased by 159.36%, and a lower distortion of simulation results than those of regular grid CA (Grid-CA). These results prove that LNU-CA can simulate land-use change better than traditional Grid-CA, and LNU is a better choice to determine the cellular unit when using CA to simulate land-use change. [ABSTRACT FROM AUTHOR]