Presentation on theme: "Comparison of Three Secondary Organic Aerosol Algorithms Implemented in CMAQ Weimin Jiang*, Éric Giroux, Dazhong Yin, and Helmut Roth National Research."— Presentation transcript:
Comparison of Three Secondary Organic Aerosol Algorithms Implemented in CMAQ Weimin Jiang*, Éric Giroux, Dazhong Yin, and Helmut Roth National Research Council of Canada
2 Outline SOA calculation in CMAQ The three CMAQ SOA algorithms Model set-up Impact on organic aerosol modelling results: spatial, temporal, SOA/fine ratios, algorithm correlations Impact on organic aerosol modelling performance: comparison with measurements Conclusions and discussion
3 SOA calculation in CMAQ Three major steps Steps 1 and 3: Binkowski and Roselle (2003); Binkowski and Shankar (1995); US EPA (1999) Implementation details: Jiang and Roth (2003) Step 2: SOA algorithm to calculate SOA mass formation rate.
4 Three CMAQ SOA algorithms Pandis: constant AYs for 6 pseudo SOA precursor species Odum: AYs for 4 pseudo species from Schell: system of equations for 10 condensable species derived from 6 pseudo species, with T correction for gas phase saturation concentrations
5 Model set-up: the model Base model: CMAQ 4.1 Modularized AERO2 by NRC (Jiang and Roth, 2002) Schell extracted from AERO3 in CMAQ 4.2 and converted to a submodule in AERO2 Three CMAQ executables: different only in SOA submodule; all other science and code the same
6 Modularized aerosol module
7 Model set-up: domain, period, inputs Nested LFV domain, Pacific ’93 episode (July 31 – August 7, 1993): see H. Roth’s presentation All model inputs are the same except for organic aerosol species: –clean IC and BC for the study of algorithm impact on modeling results –observation-base IC and BC for the study of algorithm impact on model performance
8 Impact on spatial distribution
9 Impact on temporal variation
10 Impact on model performance
11 Conclusions and discussion Schell Pandis Odum Science best among three simplified not usable SOA-generation n x Pandis 10 n x Odum very low performance good on average underestimate dramatic underestimate Note wide range of norm.bias Deficiency/problem no partitioning of org. OAY, not IAY aerosol to gas phase overestimate SOA (corrected in CMAQ 4.3 ? )
12 Odum algorithm problem: OAY vs. IAY OAY = Overall AY = average AY from ROG=0 and M 0 =0 to ROG= ROG * and M 0 =M 0 * IAY = Instantaneous AY = AY at ROG * and M 0 *
13 OAY equation vs. IAY equation Jiang (2003), Atmos. Environ. (in press)
14 OAY or IAY: A big deal? Yes, a big deal both conceptually and quantitatively.
15 Acknowledgment US EPA: Original Models–3/CMAQ Environment Canada Pollution Data Branch, Air Quality Research Branch, Pacific & Yukon Region: Raw emissions and ambient measurement data Dr. D. G. Steyn of the University of British Columbia: Pacific ’93 data set Program of Energy Research and Development (PERD) in Canada: Funding support