中文摘要：

目前中国城市化进程正处于滞后与低效并存的矛盾局面，合理的城市建设用地规模确定就成为一个焦点。但是，当前对城市建设用地规模的直接预测和间接预测两类方法都不能很好地处理快速城市化所带来的非线性问题。本文以快速城市化进程中的部分确定性信息为基础，从目标入手而不再基于历史数据的趋势外推，创建了一种目标导向的、多尺度的城市建设用地规模预测方法。这种方法包括预测区域总人口、确定城镇体系结构参数和确定城市人口规模与城市建设用地规模之间关系的三大步骤。基于1997～2008年中国各地级市的数据，构建了非农业人口与城市建设用地之间的截面和面板模型，并运用GLS等估计方法对模型中的自相关和异方差问题进行了修正，得到一个经验方程。尽管该方程在推广应用中仍存在一定局限，但相对单纯的人均用地指标更符合现实要求；以驻马店市为例对这种方法的验证也表明，该模型具有可推广的潜力。

英文摘要：

In the ongoing great-leap-forward rush of many cities, there are seemingly paradoxical co-existence of lags of urbanization and waste of land. It is not acceptable whether to take laisser-faire attitudes and make loose limitations of urban sprawl with connivance or to follow the prescribed order of urbanization without the consideration of requirements of rapid development and regional equity of cities in disadvantaged provinces, which makes it in urgent needs to find a simple and practical method which is adapted to the forecast of the urban land use size in a nonlinear process of urbanization. However, the direct and mediate methods in the mainstream of current researches on forecast of appropriate land use for urban construction cannot resolve this problem very well in great-leap-forward de- velopment of many cities, inevitably making the expansion of these cities out of order and control from central level invalid. We argued that it is required not only to improve the forecast method but also to renew the way of resolution. In furtherance of this purpose, this paper tries to explore a new multi-scalar forecasting method based on a preset goal of urbanization level from the perspective of urban system from an aim-oriented rather than a trend-extropolation perspective. This process includes three steps, first forecast the total population at the higher scale of the target city, then predict the future urban population based on the prospective urban system, at last, determine the quantity of proper urban land use through the correlation between urban land use and population. Using the data of 287 Chinese prefecture-level cities from 1997 to 2008, the basic hypothesis is tested and some panel data models on built-up areas and nonagricultural populations are built. To e- liminate the autocorrelations and heteroscedastcity in the models, some advanced methods of estimation such as GLS are introduced and an ideal empirical equation is obtained. Al- though there are still some defects and limitations of this equation