Annual Mean PM2.5 Components (EC, NH4, NO3, OC, SO4) 50m Urban and 1km Non-Urban Area Grids for Contiguous U.S., 2000-2019 v1

The Annual Mean PM2.5 Components (EC, NH4, NO3, OC, SO4) 50m Urban and 1km Non-Urban Area Grids for Contiguous U.S., 2000-2019, v1 data set contains annual predictions of the chemical concentrations at a hyper resolution (50m x 50m grid cells) in urban areas and at a high resolution (1km x 1km grid cells) in non-urban areas for the years 2000 to 2019. Particulate matter with an aerodynamic diameter less than 2.5 �m (PM2.5) increases mortality and morbidity. PM2.5 is composed of a mixture of chemical components that vary across space and time. Due to limited hyperlocal data availability, less is known about health risks of PM2.5 components, their U.S.-wide exposure disparities, or which species are driving the biggest intra-urban changes in PM2.5 mass. The national super-learned models were developed across the U.S. for hyperlocal estimation of annual mean elemental carbon, ammonium, nitrate, organic carbon, and sulfate concentrations across 3,535 urban areas at a 50m spatial resolution, and at a 1km resolution for non-urban areas from 2000 to 2019. Using Machine-Learning models (ML), combined with either a Generalized Additive Model (GAM) Ensemble Geographically-Weighted-Averaging (GAM-ENWA) or Super-Learning (SL) and approximately 82 billion predictions across 20 years, hyperlocal super-learned PM2.5 components are now available for further research. The overall R-squared values of 10-fold cross validated models ranged from 0.910 to 0.970 on the training sets for these components, while on the test sets the R-squared values ranged from 0.860 to 0.960. Remarkable spatiotemporal intra-urban and inter-urban variabilities were found in PM2.5 components. The Coordinate Reference System (CRS) for predictions is the World Geodetic System 1984 (WGS84) and the Units for the PM2.5 Components are �g/m^3. The data are provided in RDS tabular format, a file format native to the R programming language, but can also be opened by other languages such as Python.

《美国连续地区年度平均PM2.5组分（EC、NH4、NO3、OC、SO4）50米城市与1公里非城市区域网格数据集（2000-2019，版本1）》包含了对城市区域以超分辨率（50米×50米网格单元）和非城市区域以高分辨率（1公里×1公里网格单元）进行年度化学浓度预测的数据。直径小于2.5微克每立方米（PM2.5）的颗粒物质会增加死亡率和发病率。PM2.5由多种化学成分混合组成，这些成分在空间和时间上均存在差异。由于超局部数据可用性有限，对PM2.5组分对健康风险的认识较少，其在美国范围内的暴露差异，或哪些物种驱动了PM2.5质量在城市内部的巨大变化，均所知不多。为在2000年至2019年期间对3,535个城市区域的年度平均元素碳、铵、硝酸盐、有机碳和硫酸盐浓度进行超局部估计，开发了覆盖全美的国家超级学习模型，针对非城市区域采用1公里分辨率。利用机器学习模型（ML），结合广义加性模型（GAM）集成地理加权平均（GAM-ENWA）或超级学习（SL）技术，以及对20年内约820亿个预测的综合，现在可提供超局部超级学习PM2.5组分，以供进一步研究。10折交叉验证模型的总体R平方值在训练集上对这些组分介于0.910至0.970之间，而在测试集上R平方值介于0.860至0.960之间。在PM2.5组分中发现了显著的时空城市内部和城市间变异性。预测的坐标参考系统（CRS）为1984年世界大地测量系统（WGS84），PM2.5组分的单位为每立方米微克。数据以RDS表格格式提供，这是R编程语言的本地文件格式，但也可由其他语言如Python打开。