Exploring land use change is crucial to planning land space scientifically in a region. Taking the ecological conservation area (ECA) in western Beijing as the study area, we employ ArcGIS 10.2, landscape pattern index and multiple mathematical statistics to explore the temporal and spatial variation of land use from 2000 to 2020. Patch-generating Land Use Simulation (PLUS), Future Land Use Simulation (FLUS) and Markov models were used to simulate and predict the current land use in 2020. The models were evaluated for accuracy, and the more accurate PLUS model was selected and used to simulate and predict the potential land use in the study area in 2030 under two management scenarios. The main findings of this research are: (1) From 2000 to 2020, the construction land increased constantly, and the area of cultivated land and grassland decreased significantly. (2) For predicting the spatial distribution of land use in the study area, the PLUS model was more accurate than the FLUS model. (3) The land-use prediction of the study area in 2030 shows that the area of grassland, forest and water is approximately equal to their corresponding value in 2020, but the construction land increased constantly by occupying the surrounding cultivated land. According to this research, the continuous decrease of cultivated land in favor of increasing construction land will cause losses to the ecological service function of the ECA, which is not beneficial to the sustainable development of the region. Relevant departments should take corresponding measures to reduce this practice and promote sustainable development, particularly in the southern and western areas of the ECA where there is less construction land.

Over the last several decades, China has taken multiple measures for afforestation and natural forest protection, including setting the goal of carbon neutrality by the middle of 21th century. In order to support the practice of relevant policies from the scientific perspective, it is essential to precisely estimate the carbon storage of arbor forest, as it plays an important role in the carbon cycle of ecosystems. In this study, we first used the latest four phases of national forest inventory data to investigate the variation of carbon storage for both natural and planted arbor forest in China during the covered period (1999-2018). Then we used machine leaning methods to simulate the carbon density based on various kinds of environmental factors and analyzed the contribution of each influencing factor. Our results demonstrate that the total carbon storage for arbor forest in China kept increasing over the last two decades, but this increment was mainly brought about by the continuous expansion of forest land. The gap of carbon sequestration between natural forest and planted forest showed a significant trend of reduction. Additionally, tree age was identified as the dominant factor for influencing the spatiotemporal variation of carbon density among all the independent variables while the impact of climatic factor was limited. Therefore, the future improvement of carbon sequestration of arbor forest in should mainly rely on additional projects of afforestation, reforestation, green space conservation and reduction of emissions in China. Conclusions of this study have important implications for policy makers and other stakeholders to evaluate the previous achievement of environmental projects and can also help to set future plans and finally realize the goals of carbon neutrality.

As more machine learning and deep learning models are applied in studying the quantitative relationship between the climate and terrestrial vegetation growth, the uncertainty of these advanced models requires clarification. Partial dependence plots (PDPs) are one of the most widely used methods to estimate the marginal effect of independent variables on the predicted outcome of a machine learning model, and it is regarded as the main basis for conclusions in relevant research. As more controversies regarding the reliability of the results of the PDPs emerge, the uncertainty of the PDPs remains unclear. In this paper, we experiment with real, remote sensing data to systematically analyze the uncertainty of partial dependence relationships between four climate variables (temperature, rainfall, radiation, and windspeed) and vegetation growth, with one conventional linear model and six machine learning models. We tested the uncertainty of the PDP curves across different machine learning models from three aspects: variation, whole linear trends, and the trait of change points. Results show that the PDP of the dominant climate factor (mean air temperature) and vegetation growth parameter (indicated by the normalized difference vegetation index, NDVI) has the smallest relative variation and the whole linear trend of the PDP was comparatively stable across the different models. The mean relative variation of change points across the partial dependence curves of the non-dominant climate factors (i.e., radiation, windspeed, and rainfall) and vegetation growth ranged from 8.96% to 23.8%, respectively, which was much higher than those of the dominant climate factor and vegetation growth. Lastly, the model used for creating the PDP, rather than the relative importance of these climate factors, determines the fluctuation of the PDP output of these climate variables and vegetation growth. These findings have significant implications for using remote sensing data and machine learning models to investigate the quantitative relationships between the climate and terrestrial vegetation.