本文在成本效益理论和概率理论的基础上,以武汉市洪山区为例,建立Probit模型研究农地城市流转概率,并预测其空间分布。为保证样本的随机性和避免空间相关性对概率模型的估计误差,采用800m×800m的格网将研究区域格网化,并将格网中心作为研究样点。模型中引入了样点自然特征变量、区域经济因素变量和政策措施变量,通过读取278个样点的GIS属性值和统计年鉴数据值进行回归分析。回归结果表明:①农地周围建设用地比例、规划限制对农地城市流转概率有显著影响,离河流、湖泊和绿地的距离及地块单位面积GDP对流转概率也有较大影响;②研究区域高流转概率区面积约为41,36hm^2,中高概率区面积约为371,42hm^2;③新村村、关山村、钢铁村、先建村等处的基本农田流转概率较高,是流转管控重点区域;武丰新村、南湖村、渔光村、青菱村、渔业村等区域则应在符合数量控制指标下优先流转。根据理论和实证研究结论,城市在制定土地利用规划时,应综合考虑地块的区位特征和土地利用的空间相关性,并重视土地规划与农地城市流转概率相互联系和相互影响,以利于规划更为合理和切合实际。
The spatial pattern and distribution of land use has important environmental consequences. Based on cost-benefit analysis and probability theory, a probit model was used to explore the probability of rural-urban land conversion and its spatial distribution in Hongshan District, Wuhan. To randomize the sample and diminish the effects of spatial correlation, an 800m by 800m square was used to divide the research area into 278 grid cells, and the central points of these cells were used as sample points for estimating agricultural land conversion probability. The model included variables for parcel attributes, economic characteristics and land conversion policies, based on GIS data and official statistics. The results of a regression analysis showed that probability of agricultural land conversion was significantly affected by neighboring land use. In particular, when the proportion of developed land in a parcel increased, the conversion probability of remaining undeveloped land in that parcel increased correspondingly. Urban planning also significantly affected the probability of land conversion. If agricultural land was located in an urban planning area, the probability of urban development was more than 90 %. Distances to the Yangtze River, large lakes and parks also influenced conversion probability. This study identifies villages with high conversion probability where prime farmland should be strictly protected, and other villages where preferential development policies should be targeted. We suggest that local governments should adequately consider site-specific attributes, spatial correlation among neighboring parcels, and the relationship between planning and rural-urban conversion probability in the land use planning process.