Table 2.
PM2.5 data | AQS+IMPROVE | CDC WONDER | EPHTN | Dalhousie |
---|---|---|---|---|
Minimum PM2.5 value (μg/m3) | 1.7 | 7.2 | 4.4 | 3.3 |
Maximum PM2.5 value (μg/m3) | 20.4 | 14.9 | 16.0 | 13.4 |
R2 * | 0.106** | 0.649 | 0.647 | |
R2 * for PM2.5 < 15 μg m−3 only | 0.136** | 0.649 | 0.662 | |
Normalized mean bias * | 33.3%** | 12.2% | −3.3% | |
Number of samples used in analysis | 544 | 3102 | 3104 | 3104 |
R2 and normalized mean bias are calculated based on county-mean to county-mean comparison with respect to AQS+IMPROVE data of 2011 annual mean.
This is based on county-mean to county-mean comparison, but as for the grid-level validation, the surfacing algorithm that was used to create the CDC WONDER dataset was also validated on the grid level for the Southeastern US that showed an R2 of 0.88 and RMSE of 1.6 μg/m3 (Al-Hamdan et al., 2009).