Fractional reaction–diffusion modeling and machine learning for vegetation pattern analysis in Junggar Basin under climate change

This study investigates the mechanistic effects of vegetation physiological processes and develops a refined vegetation-climate dynamic model with a fractional-in-space diffusion model. The model comprehensively integrates key climatic factors, such as precipitation, temperature, and CO2, to examine the impact of climate change on the evolution of vegetation patterns in the Junggar Basin. Through analysis, we find an inverse relation between the fractional-order coefficient and the size of the Turing instability domain. In addition, performing numerical simulations using real data from the Junggar Basin region, the results show that the interaction between heat stress and the effect of water and CO2 fertilization significantly affect vegetation growth. What is more, the future vegetation growth under different climate scenarios is predicted based on the current scenario and three climate scenarios from the Coupled Model Intercomparison Project Phase 6. We harness the predictive capabilities of machine learning algorithms to forecast changes in the current scenarios. The numerical results show that the current and the SSP1-2.6 scenarios are the favorable climate scenario for vegetation growth. In contrast, the SSP2-4.5 and SSP5-8.5 scenarios suppress vegetation growth and the SSP5-8.5 scenario exhibits the fastest rate of desertification.