Study on the Gridded GDP Based on Oasis LandUse:A Case of the northern slope of Tianshan Mountains

随着对全球变化过程的深化研究，人类活动与地球环境变化间的相互作用机制受到越来越多的关注，社会经济数据对于全球、洲际以及区域尺度上人与环境交互作用研究的重要性已被广泛认知。然而传统以行政区划为单位进行表示的社会经济数据和以自然单元为单位的环境数据在空间单元上的不匹配使得人与环境的交互作用分析变得十分困难。特别是在我国广大的西部干旱地区，县域范围内往往包括广袤的沙漠和戈壁，人类活动主要集中在水土资源条件较好的绿洲之上，社会经济数据的传统表示方法无法真实反映人类活动的空间分布状况。因此，对社会经济数据进行空间化，通过建立模型将经济发展水平落实到栅格层次的空间上，模拟社会经济的空间分布状况和动态变化过程成为当前研究的热点。 本研究选择新疆天山北坡为研究区，该区域从南向北依次分布着山地－绿洲－荒漠生态系统。其中，山地和荒漠地区由于自然条件的限制，人类活动进行的难度较大，而中部的绿洲区域水土条件相对较好，再加上丰富的自然资源，人类活动强度较大。天山北坡经济的发展具有典型的绿洲经济特色，县域范围内人口和经济要素的空间分布极不均匀，对于干旱区经济的空间分布具有很好的代表性。本文在分析干旱区经济发展特征的基础上，根据分县控制、分产业建模的思路，建立基于土地利用的1 km 栅格GDP 空间化模型对统计型经济数据进行空间分布仿真模拟。通过研究得到以下主要结论： （1）本研究确立的分县控制、分产业建模、以土地利用为基础的建模原则综合考虑了干旱区绿洲经济特征，避免了不同地区以及同一地区不同地类上GDP 的再分配，保证了模拟模型的正确性和普适性。 （2）干旱区第一产业主要依赖于耕地、草地、林地和水域等地类，其面积和土地质量等直接决定着一产产值的空间分布；第二产业的发展与道路密切相关，二者间本文由国家973 重点研究项目（2009CB421301），国家科技支撑计划项目（2007BAH12B03）和新疆主体功能区划项目联合支持。 II 基于土地利用的干旱区GDP 公里格网化研究——以新疆天山北坡为例的拟合关系符合距离衰减规律；第三产业的发展与城市的发展进程紧密联系，城镇规模越大，第三产业越发达，城镇内部三产产值的空间分布表现出从城市中心向四周辐射递减的规律。 （3）针对各产业不同的影响因素，本文对第一产业产值采用面积权重的方法进行拟合；对于第二产业，建立基于道路的反距离加权模型；第三产业产值则以城镇规模为基础，建立距离衰减加幂指数模型，并引入边缘距离概念以表征大城市多中心发展的情况。 （4）通过在分产业和分县两个尺度上对研究区GDP 空间分布模拟结果的精度检验，并与全国尺度GDP 空间分布模拟结果进行对比分析，结果表明，对于天山北坡GDP 的空间分布模拟精度较高，本研究采用的方案和模型在一定程度上弥补了前人的不足，具有一定的合理性。 （5）本研究所得的各产业产值及GDP 的空间分布模拟结果与研究区经济发展的实际情况相一致，较为真实地反映了干旱区串珠状经济发展特征，建成区内部以及建成区与周边乡镇的经济差异也得以较好地体现。 文章所采用的方法综合考虑了土地利用、道路和城市规模对经济的影响，避免了GDP 在各地类内部平均化分布的情况，较为真实地反映了区域经济的空间分布特征，为社会经济数据与自然地理要素数据的有效结合做出了新的尝试，对准确把握区域经济动态发展具有一定的实用价值和现实意义。It is widely recognized that socio-economic data contribute to integrative research on the dynamic interaction between nature and society in regional, continental and global scale. The reciprocal relationship between economy and human activities were documented by many researchers. The economic development is inseparable from the participation of human beings, meanwhile, the population increasing will somehow increase resource consumption and degrade the ecosystem functioning. Generally, the socio-economic data is derived from the administrative regions at county or province level in China, while the environmental data reduction is based on the natural units. The difference of spatial units between these two kinds of data led to uncoordination of integrative research among society, economy and environment. The area of desert and Gobi are included in county in arid regions, located in western China, and human activities are concentrated in the oasis area. As a result, the analysis using traditional methods with economic data can not represent the spatial distribution of human activities. Therefore, modeling and simulating the spatial dynamic change of economic value on the basis of socio-economic data spatialization in Grid-cells is useful method to understand the distribution of human activities, and which has become a hotspot in interdisplinary and transdisplinary research. The northern slope of Tianshan Mountains located in Western China, is a rich ecosystem biodiversity area. The mountainous, oasis, and wilderness ecosystems are distributed in turn from south to north in this area. Due to the restriction of water, soil and other nature resources, the human activities are mainly congregated in oasis region, which resulted in the unusual distribution of population and economic development within a county. Through analyzing the character of the region economic development, we consider the GDP of counties separately and developed the model based on land use for three types of industries to simulate the difference of GDP in arid area using 1 square kilometers grid-cell. We took the GDP data in the northern slope of Tianshan Mountains in Xinjiang, in 2000 as a case, the main findings are as follows: (1) The idea of modeling which has based on land use conditions, took industries into account respectively and set the economic level as a control variable was built. It has considered the character of oasis economy in arid regions, and avoided the redistribution of GDP in different land categories. So the validity and the generality of the models are guaranteed. (2) The development of primary industry is depended on cropland, grass land, forest land and water area in arid regions. The spatial distribution of first industrial output is determined by the area and productivity of these land types. The development of second industry is significant related to the road density in cities, and comply with the distance decay rule. In terms of tertiary industry, its development is relied on the process of urban development. A positive correlation existed between urban scale and the development of tertiary industry. Within a city area, the tertiary industry output value is decreased from the urban center to suburban. (3) We developed three different models to simulate the spatial distribution of there kinds of industries because of the difference of factors. To the primary industry, we adopt the area weightiness method, using linear equation to fit it; According to the influence of road on secondary industry, build road-based counter-distance weighted model to calculate the secondary industry output indices, and then implement the spatialization of the secondary industry output. For the tertiary industry, a power exponential model based on the urban scale and the distance from the urban center, derived from distance decay function, and took the distance from urban edge into power exponential model to simulate the spatial distribution of tertiary industry output value in big cities. (4) The results of accuracy assessment indicated that the precision of each simulation result is highly both at industry and county level in the northern slope of Tianshan Mountains. The simulation method used in this paper is more practicable and effective than traditional methods through comparing with the typical GDP estimation results of cngdp2000 in country scale. (5) The gird maps of output values per square km of each industry and GDP were highly consistent with the actual conditions of study area. In the northern slope of Tianshan Mountains, the development of economical follows the points-axis pattern. The simulation results reflected the real distribution characteristics of regional economy, and the built-up district interior as well as the built-up district and the peripheral villages’ economical differences are also manifest well. In this paper, we took land use, road and urban scale into account to simulate the distribution of economical. The models we built can reflected the real distribution characteristics of arid regions, and avoid the situation that traditional method equally distribute GDP in the same land category. It is a new attempt to integrate the socio-economic data with the physical geography data together, and the model practicable and effective to understand the spatial dynamics of the economic development and the distribution of human activities in arid regions