Simulating spatiotemporal land use change in middle and high latitude regions using multiscale fusion and cellular automata: The case of Northeast China

Abstract

Global warming-induced land use change in middle and high latitude regions remains an important and challenging topic for sustainable development. Although a large amount of remote sensing imageries is available for extracting land use information, there are unavoidable inconsistencies among land use products from different platforms. This study proposes a framework which integrates multiscale fusion method and cellular automata to simulate spatiotemporal land use change in middle and high latitude regions. Five types of remote sensing-based land use products were used to extract land use information of Northeast China in 2001, 2009, and 2017. The results demonstrate that the fused land use product from the integration of remote sensing-multiscale fusion and cellular automata has a higher accuracy. The simulated land use map for 2017 using CA-Markov model was validated by the actual land use classification of 2017, and the stochastic Kappa (Kno), location related Kappa (Klocation) and traditional Kappa (Kstandard) were 0.90, 0.94 and 0.86, respectively. The results of land use change from 2001 to 2017 show the biggest increase (5.9%) in farmland, mainly from grassland and forest. The predicted results indicate that this trend will persist until 2025 with a continuous augmentation of farmland in Northeast China. The areas of grassland will decrease by 4.3% from 2017 to 2025, with a major conversion to farmland and forest. The increasing farmland is a potential threat to ecology environment in the region as it will heavily sacrifice grassland and forest. These findings provide useful information on understanding spatiotemporal land use change in middle and high latitude regions and allow us maintaining food security for sustainable development.