Implementation and application of sub-grid-scale plume treatment in the latest version of EPA’s third-generation air quality model, CMAQ 5.01

Figures & data

Figure 1.  Modeling domains (left panel) for PinG simulations. The outer 36-km domain provides the boundary conditions for the inner 12 km domain used in the simulations. The right panel shows the locations of the point sources treated with PinG.

Table 1.  Stack parameters and emissions from point sources selected for PinG treatment

Location		Stack parameters					Emissions (tons per year)
Latitude	Longitude	Diameter	Flow rate	Height	Temperature	Velocity
		(m)	(m^3/s)	(m)	(K)	(m/s)	NOx	PM2.5	PM10	SO2	CO
38.3579	−76.9788	6	831	213	405	30	6737	896	1554	38553	270
41.8912	−83.3450	9	2232	213	405	39	4505	434	471	14672	333
41.8907	−83.3464	9	2232	213	403	39	5600	392	416	15666	307
36.2825	−80.0592	7	1829	183	426	42	10888	2151	2470	49083	678
38.6361	−83.6939	6	850	244	411	32	7185	1300	1452	29983	428
38.8906	−82.1271	9	2102	305	450	33	3886	968	1094	15059	160
38.9356	−82.1153	13	2081	253	325	16	20331	1118	1278	14998	1050
39.8555	−79.9268	7	1549	213	428	42	6167	3099	4037	53861	319
40.3842	−79.0611	9	1232	160	321	22	10108	934	1750	3921	702
39.3836	−80.3161	8	827	305	339	17	7177	673	992	1699	438
39.8325	−80.8130	10	2616	366	429	33	12046	1019	1073	31592	399
38.4728	−81.8231	9	2020	275	436	30	8310	1173	1242	31519	503
34.1256	−84.9192	8	944	305	403	21	7153	2337	3093	48001	546
34.1256	−84.9192	8	1236	305	410	27	7151	2725	3839	50307	582
33.4083	−85.0331	8	1133	305	430	25	8191	2346	2830	55284	671
33.0586	−83.8069	8	1232	305	410	23	4924	1251	1404	22384	679
33.0611	−83.8064	8	1280	305	401	24	4485	364	390	22428	925
38.7397	−85.4203	7	1266	300	450	34	4591	132	221	13323	201

Table 2.  Model performance statistics for July 1 to 15, 2005, 1-hr average ozone mixing ratios, using 40 and 60 ppb cutoffs on observed ozone concentrations

	40 ppb cutoff		60 ppb cutoff
Metric	CMAQ base	CMAQ APT	CMAQ base	CMAQ APT
Mean observed value (ppb)	55.4	55.4	70.2	70.2
Mean predicted value (ppb)	53.9	53.9	63.6	63.6
Ratio of means	0.97	0.97	0.91	0.91
Gross bias (ppb)	−1.5	−1.5	−6.6	−6.6
Normalized bias (%)	−1.5	−1.4	−9.1	−9.1
Fractional bias (%)	−4.4	−4.4	−11.3	−11.2
Gross error (ppb)	9.4	9.4	10.3	10.2
Normalized error (%)	17.4	17.4	14.5	14.5
Fractional error (%)	18.4	18.4	16.2	16.1
Normalized mean bias (%)	−2.7	−2.7	−9.4	−9.4
Normalized mean error (%)	16.9	16.9	14.6	14.6
Coefficient of determination (r^2)	0.30	0.30	0.19	0.19
Number of points	49765	49765	14397	14397

Table 3.  Model performance statistics for January 1 to 15 and July 1 to 15, 2005, 24-hr average PM_2.5 concentrations

	January 2005		July 2005
Metric	CMAQ base	CMAQ APT	CMAQ base	CMAQ APT
Mean observed value (μg/m^3)	7.7	7.7	12.8	12.8
Mean predicted value (μg/m^3)	11.2	11.2	8.7	8.7
Ratio of means	1.45	1.45	0.68	0.68
Gross bias (μg/m^3)	3.5	3.5	−4.0	−4.1
Normalized bias (%)	51.9	52.0	−32.9	−33.1
Fractional bias (%)	23.2	23.5	−48.8	−49.0
Gross error (μg/m^3)	4.8	4.8	4.9	4.9
Normalized error (%)	70.3	70.1	40.2	40.2
Fractional error (%)	47.5	47.2	54.7	54.6
Normalized mean bias (%)	44.9	45.0	−31.6	−31.8
Normalized mean error (%)	61.6	61.5	38.6	38.5
Coefficient of determination (r^2)	0.52	0.52	0.55	0.56
Number of points	171	171	137	137

Figure 2.  Locations of AQS stations within a 9 by 9 (108 km by 108 km) grid cell region around each PinG source.

Table 4.  Model performance statistics for July 1 to 15, 2005, 1-hr average ozone mixing ratios at AQS monitors within a 5 by 5 grid cell domain around each PinG point source, using 40 and 60 ppb cutoffs on observed ozone concentrations

	40 ppb cutoff		60 ppb cutoff
Metric	CMAQ base	CMAQ APT	CMAQ base	CMAQ APT
Mean observed value (ppb)	55.1	55.1	70.7	70.7
Mean predicted value (ppb)	55.4	55.2	64.5	64.2
Ratio of means	1.01	1.00	0.91	0.91
Gross bias (ppb)	0.3	0.1	−6.2	−6.5
Normalized bias (%)	2.1	1.8	−8.3	−8.7
Fractional bias (%)	−1.3	−1.6	−10.5	−10.9
Gross error (ppb)	10.0	9.9	10.2	10.2
Normalized error (%)	19.0	18.7	14.2	14.2
Fractional error (%)	19.6	19.4	15.9	15.8
Normalized mean bias (%)	0.5	0.1	−8.8	−9.2
Normalized mean error (%)	18.2	18.0	14.4	14.4
Coefficient of determination (r^2)	0.22	0.22	0.10	0.11
Number of points	2791	2791	750	750

Table 5.  Model performance statistics for July 1 to 15, 2005, 1-hr average ozone mixing ratios at AQS monitors within a 9 by 9 grid cell domain around each PinG point source, using 40 and 60 ppb cutoffs on observed ozone concentrations

	40 ppb cutoff		60 ppb cutoff
Metric	CMAQ base	CMAQ APT	CMAQ base	CMAQ APT
Mean observed value (ppb)	56.1	56.1	70.3	70.3
Mean predicted value (ppb)	55.7	55.6	64.9	64.8
Ratio of means	0.99	0.99	0.92	0.92
Gross bias (ppb)	−0.4	−0.5	−5.4	−5.5
Normalized bias (%)	0.7	0.5	−7.3	−7.4
Fractional bias (%)	−2.5	−2.6	−9.1	−9.2
Gross error (ppb)	9.7	9.7	9.4	9.3
Normalized error (%)	18.2	18.1	13.2	13.1
Fractional error (%)	18.9	18.8	14.6	14.5
Normalized mean bias (%)	−0.8	−0.9	−7.7	−7.8
Normalized mean error (%)	17.4	17.2	13.4	13.3
Coefficient of determination (r^2)	0.27	0.27	0.14	0.15
Number of points	7197	7197	2276	2276

Figure 3.  Time series of hourly surface ozone mixing ratios at two AQS monitoring sites in a 108 by 108 km region around PinG sources. The Morgantown, WV, site is about 23 km south of the nearest source, while the Holbrook, PA site is about 31 km west of the nearest source.

Figure 4.  Predicted source contributions to maximum daily 8-hr average surface ozone mixing ratios (ppb) in the first model layer (center at about 20 m from ground level) from the base CMAQ (left panel) and CMAQ-APT (middle panel) and differences between the APT and base model predicted ozone concentrations (right panel) for July 11, 2005 (top panels), and July 12, 2005 (bottom panels). The circles in the plots represent the point sources selected for PinG treatment.

Figure 5.  Predicted source contributions to 24-hr average surface PM_2.5 concentrations (μg/m³) in the first model layer (center at about 20 m from ground level) from the base CMAQ (left panel) and CMAQ-APT (middle panel) and differences between the APT and base model predicted PM_2.5 concentrations (right panel) for January 9, 2005 (top panels), and January 15, 2005 (bottom panels). The circles in the plots represent the point sources selected for PinG treatment.

Figure 6.  Predicted source contributions to 24-hr average surface PM_2.5 concentrations (μg/m³) in the first model layer (center at about 20 m from ground level) from the base CMAQ (left panel) and CMAQ-APT (middle panel) and differences between the APT and base model predicted PM_2.5 concentrations (right panel) for July 11, 2005 (top panels), and July 12, 2005 (bottom panels). The circles in the plots represent the point sources selected for PinG treatment.

Figure 7.  Predicted source contributions to 15-day total sulfur deposition (eq/ha) from the base CMAQ (left panel) and CMAQ-APT (middle panel) and differences between the APT and base model predicted total sulfur deposition (right panel) for January 2005 (top panels) and July 2005 (bottom panels). The circles in the plots represent the point sources selected for PinG treatment.