基于Logit 模型的世界主要作物播种面积变化模拟

农作物播种面积动态变化因其重要性已经引起了国内外学者的广泛关注.从"人-地关系"中人的角度出发,利用多元Logit模型初步建立了全球尺度的农作物播种面积变化模拟系统,分析研究了未来30年内世界主要农作物播种面积变化的数量特征和空间格局.模型建立思路是:作物播种面积变化是农户作物选择行为的直接结果,而选择何种作物进行播种是由作物效用决定的.因此,利用离散选择理论,选择影响作物效用大小的主要解释变量建立效用函数,动态模拟农户作物选择行为,并得到这种选择行为所带来的农作物播种面积变化的时空特征.模型建立后,利用联合国粮农组织(FAO)2001-2003年统计数据和2001年MODIS全球土地覆盖数据产品对模型结果进行验证,结果表明模型运行可靠,和实际状况吻合较好,可以应用于未来情景模拟分析.其后,以5年为步长,模型对世界四大作物(水稻、玉米、小麦和大豆)在2005-2035年间的播种面积动态变化进行了模拟,从全球作物总播种面积变化差异、四大作物播种面积变化的区域差异和不同区域内四大作物播种面积变化差异等方面进行了分析研究,结果表明:不同作物播种面积变化的数量特征和空间格局是不相同的.虽然目前模型还存在一些不确定性,但仍然能够在一定程度上为理解现在和未来农业土地利用的复杂动态变化提供帮助,模拟结果可为有关部门提供决策支持和信息服务.

Global-scale Modeling of Future Changes in Sown Areas for Major Crops Based on a Logit Model

Potential dynamics of agricultural crop sown area has attracted a wide range of attention from numerous researchers due to its ecological and socio-economic implications. This paper describes a method to develop and implement an integrated model to dynamically simulate future changes in sown areas of the world's major crops (rice, maize, wheat and soybean) at a global scale. The general hypothesis of this study is that crop sown area change is directly linked with farmer decisions on crop choice, and what kind of crop to be cultivated is highly dependent on the random utility of available crops. The modeling approach is based on a crop choice model, which is a Multinomial Logit Model and used to model farmer crop choice decisions among a variety of available alternatives by using an optimization approach. The assessment of model performance by comparing model estimates with FAO statistical data (2001-2003) and MODIS land cover product (2001) indicates its reliability and dignity for addressing the complicated dynamic change of agricultural land use change at present and capability for long-term scenario investigation and applications for the future. From model simulation for crop sown area change during 2005-2035 in different regions in the world, global potential cropping patterns of major crops can be interpreted. Moreover, the results present that the change rates and trajectories of crops in different regions show a great variation over time and space. This study is an attempt of detecting future sown area change at a global level by using a simplified approach along with some assumptions. Although some uncertainty remains in the model, the outcomes can help to understand and explain the causes, locations, consequences and trajectories of land-use change, and provide a great support service for land-use planning and policy-making activities.