United States Geological Survey, California Water Science Center, San Diego, CA, United States.
United States Geological Survey, New England Water Science Center, Northborough, MA, United States.
Sci Total Environ. 2017 Dec 31;607-608:658-668. doi: 10.1016/j.scitotenv.2017.07.018. Epub 2017 Jul 27.
We estimate the location and population served by domestic wells in the contiguous United States in two ways: (1) the "Block Group Method" or BGM, uses data from the 1990 census, and (2) the "Road-Enhanced Method" or REM, refines the locations by using a buffer expansion and shrinkage technique along roadways to define areas where domestic wells exist. The fundamental assumption is that houses (and therefore domestic wells) are located near a named road. The results are presented as two nationally-consistent domestic-well population datasets. While both methods can be considered valid, the REM map is more precise in locating domestic wells; the REM map has a smaller amount of spatial bias (Type 1 and Type 2 errors nearly equal vs biased in Type 1), total error (10.9% vs 23.7%), and distance error (2.0km vs 2.7km), when comparing the REM and BGM maps to a calibration map in California. However, the BGM map is more inclusive of all potential locations for domestic wells. Independent domestic well datasets from the USGS, and the States of MN, NV, and TX show that the BGM captures about 5 to 10% more wells than the REM. One key difference between the BGM and the REM is the mapping of low density areas. The REM reduces areas mapped as low density by 57%, concentrating populations into denser regions. Therefore, if one is trying to capture all of the potential areas of domestic-well usage, then the BGM map may be more applicable. If location is more imperative, then the REM map is better at identifying areas of the landscape with the highest probability of finding a domestic well. Depending on the purpose of a study, a combination of both maps can be used.
（1）“街区组方法”（BGM），使用 1990 年人口普查数据；（2）“道路增强方法”（REM），通过在道路沿线使用缓冲区扩展和收缩技术来细化位置，以定义存在家用井的区域。基本假设是房屋（因此也包括家用井）位于指定道路附近。结果呈现为两个具有全国一致性的家用井人口数据集。虽然两种方法都可以被认为是有效的，但 REM 地图在定位家用井方面更为精确；与 BGM 地图相比，REM 地图的空间偏差更小（1 型和 2 型错误几乎相等，而偏向于 1 型），总误差（10.9% 对 23.7%）和距离误差（2.0 公里对 2.7 公里），当将 REM 和 BGM 地图与加利福尼亚州的校准地图进行比较时。然而，BGM 地图更全面地包含了所有潜在的家用井位置。来自 USGS 以及 MN、NV 和 TX 州的独立家用井数据集表明，BGM 比 REM 多捕捉到约 5%至 10%的水井。BGM 和 REM 之间的一个关键区别是低密度区域的映射。REM 将低密区的映射减少了 57%，将人口集中到更密集的区域。因此，如果试图捕捉家用井使用的所有潜在区域，那么 BGM 地图可能更适用。如果位置更重要，那么 REM 地图更擅长识别最有可能找到家用井的景观区域。根据研究的目的，可以使用这两个地图的组合。
