长江中游城市群土地利用转型和城镇化之间的空间关系

城市群正成为中国新型城镇化主体形态，城市群集聚区的快速城镇化对土地利用转型产生了前所未有的影响。但是不清楚的土地利用转型特征和城镇化之间的空间关系，会影响土地利用政策制定和城镇化的健康发展。该文借助1995～2015年间5期土地利用现状遥感监测数据，在5和10 km格网尺度上，从土地利用综合动态度、土地利用程度和土地利用多样性等3方面测度了长江中游城市群土地利用转型水平。借助普通最小二乘法（OLS）、空间滞后模型（SLM）、空间误差模型（SEM）和加入空间滞后项的空间误差模型（SEMLD）探测了土地利用转型和城镇化之间的空间和非空间关系。研究结果显示：（1）1995～2015年间长江中游城市群土地利用发生了深刻转型，耕地面积净减少 6 153.80 km²，建设用地面积净增加 6 205.25 km²；（2）地级市以及周边格网和主要的交通道路沿线格网土地利转型速度、土地利用程度变化和土地利用多样性变化显著高于其他地区；（3）不同格网尺度城镇化和土地利用转型之间具有显著的正向空间自相关关系；（4）回归结果显示，SEMLD模型可以更好地解释土地利用转型与城镇化之间的空间关系，在不同尺度回归分析中发现土地利用转型水平对城镇化具有显著的空间依赖性。未来的土地利用管理以及城镇化发展中需要考虑二者之间的显著的空间依赖性。

On the Spatial Relationship Between Land Use Transition and Urbanization in the Middle Yangtze River Urban Cluster

The urban agglomeration is increasingly becoming the main areas of China’s new urbanization. The rapid urbanization in urban clusters has an unprecedented impact on land use transition (LUT). However, unclear spatial relationships between LUT and urbanization may limit the sustainable land use and the healthy development of urbanization. Based on the Landsat TM/ETM image interpretation data from 1995 to 2015, this paper measures LUT in the middle Yangtze River Urban Cluster (MYRUC) in three ways: land use dynamics degree, land use intensity, and land use diversity at 5 and 10 km grid scales. Our study examines the spatial and non spatial relationships between land use transition and urbanization level using ordinary least squares model (OLS), spatial lag model (SLM), spatial error model (SEM), and spatial error model with lag independence (SEMLD). The results indicated that: (1) During 1995-2015, the land use in the MYRUC had undergone a profound transition. The net decrease of cultivated land was 6 153.80 km² and the net increase of construction land was 6 205.25 km². (2) The land use change speed, land use intensity and land use diversity in the prefectural level municipal districts, the surrounding grids and grids along the main traffic routes were significantly higher than other grids. (3) There was a significant positive spatial autocorrelation between urbanization and LUT at different grid scales. (4) The regression results showed that the SEMLD model could better explain the spatial relationship between LUT and urbanization than other models. Significant spatial dependence was found between LUT and urbanization in both 5 and 10 km grid scales. Our results suggest that policy makers should consider the significant spatial dependence between LUT and urbanization in the future land use management and urbanization development.