Utah Buildings

SGID.LOCATION.Buildings was derived from building footprints generated by Microsoft for all 50 States https://github.com/Microsoft/USBuildingFootprints In some cases the pixel prediction algorithm used by Microsoft identified and created building footprints where no buildings existed. To flag potential errors, building footprints within 750 meters of known populated areas (SGID.DEMOGRAPHIC.PopBlockAreas2010_Approx) and within 500 meters of an address point (SGID.LOCATION.AddressPoints) were selected and indentified as being a likely structure, footprints falling outside these areas were identified as possible buildings in the 'TYPE' field. Additional buildings were then added from OpenStreetMap data, when they were more than three meters from an already-existing building in the dataset. Finally, attributes were added for address, city, county, and ZIP code where possible.SGID.LOCATION.Buildings 数据集源自微软为全美50个州生成的建筑轮廓（building footprints）数据，原始数据仓库地址为 https://github.com/Microsoft/USBuildingFootprints。微软所使用的像素预测算法有时会在无建筑的区域识别并生成建筑轮廓。为标记潜在错误，我们选取了两类建筑轮廓：一类是位于已知人口聚居区（SGID.DEMOGRAPHIC.PopBlockAreas2010_Approx）750米范围内的轮廓，另一类是距离地址点（SGID.LOCATION.AddressPoints）500米范围内的轮廓，将其认定为疑似真实建筑；落在上述范围之外的轮廓则在"TYPE"字段中被标注为疑似建筑。随后，我们从开放街道地图（OpenStreetMap）数据中补充了额外建筑，要求这些新增建筑与数据集中已有的建筑间距超过3米。最后，在条件允许的情况下，为数据集补充了地址、城市、县及邮政编码等属性字段。