Environmental Engineering Reference
In-Depth Information
of each grid cell, and then takes an inverse distance squared weighted average of
the monitor concentrations. The “fused” spatial fields are calculated by adjusting
the interpolated ambient data (in each grid cell) up or down by a multiplicative
factor calculated as the ratio of the modeled concentration at the grid cell divided
by the modeled concentration at the nearest neighbor monitor locations (weighted
by distance). Equation 1 shows the calculation of the fused concentrations.
Model
Model
n
∑
1
E baseline
,
Species
=
Weight
⋅
Monitor
⋅
E baseline
,
i
i
i
=
i baseline
,
The concentration of Species E (which can be PM2.5, ozone, or any PM2.5
species component) is calculated as the sum of the weighted concentrations at
each neighbor monitor. The multiplicative model adjustment factor, which is the
last term in the equation, is only used in the fused data fields calculation.
3. Spatial Fields Performance
In order to test the performance of the interpolated and fused interpolated spatial
fields, we employed a cross validation technique. Using MATS, we produced
several sets of fields which were developed using various sets of ambient data and
photochemical model outputs. The “control” case spatial fields were run for the
Eastern 12 km domain using all available PM2.5 ambient data for 2002. The data
was derived from the Federal Reference Method (FRM) and IMPROVE monitoring
networks. Within the domain, there were 855 sites which had complete data
2
for all 4 quarters of 2002. A set of annual average spatial fields for 2002 were
created using the data from all 855 sites. This was designated the “control case”.
Additionally, the control case interpolations were combined with photochemical
model output for 2002
3
to create a “fused control case” (using the calculations
from Eq. 1).
Table 1
shows the performance statistics for these two spatial fields.
As expected, the performance at the monitors is excellent.
4
2
Complete data was defined as having ≥11 samples per quarter.
3
The analysis used output from the CMAQ photochemical model. Documentation of the 2002
modeling platform that was used can be found here: http://www.epa.gov/scram001/reports/
Emissions%20TSD%20Vol1_02-28-08.pdf.
4
The bias and error is not zero because the data is interpolated to the center of the 12km grid
cells, and not to the actual monitor locations.
