【摘要】 以国家统计局公开发布的2005年全国1%人口抽样调查数据为基数,对全国县级尺度农村房屋结构的比例构成进行了推定。针对全国性普查数据存在的问题,开发了基于不同空间精度等级的数据收缩模型,利用中国建筑气候区划、建筑热工区划、人口和社会经济指标等因子,推定了2300多个县级空间单元的农村地区不同建筑结构的房屋比例和数量。同时,对模型结构和推定精度进行了校验,结果表明,由地级和省级单元的五种房屋结构比例来推定县级单元数据,模型精度(R2)分别达到70%和40%。这项工作为从区域尺度把握农村房屋结构状况,评价各地对地震灾害的抵御能力和制定相关规划提供了基础数据支撑。更多还原

【Abstract】 County-level housing structure data,especially those in rural areas,are critical for evaluating regional seismic vulnerability and making anti-disaster planning.The data open to the public,however,are incomplete and inconsistent in scale.This paper addressed this issue and estimated the county-level rural housing structure data across the country,including rural and urban housing number and housing ratios in five structures,including wood,brick,mixed,and reinforced concrete.The national 1/100 census data conducted in 2005 were taken as the basic data.Based on an evaluation of the original data,four spatial accuracy levels were specified according to their statistical scales.Separation method was used to identify rural and urban housing data individually from total housing data.Besides,spatial down-scaling models were established to transfer housing data from province-level or prefecture-level to county-level.The influences of geographic factors,including architectural climate zone and architectural thermo zone,and economic and social factors,for example,average relative GDP per capita,minority counties,were considered in the models,and also spatial autocorrelation was considered in prefecture-level models.Linear regression models and MLP models were compared with Rsqure and RMSE when making spatial down-scaling models,and linear regression models were adopted.Finally,models’ Rsqures reached 70% and 40% corresponding to prefecture-level and province-level models,which seemed acceptable facing the incomplete and inconsistent data at present.更多还原