Elements of a Weight of Evidence Determination Donna Kenski Midwest RPO presented at National RPO Meeting, Denver, June 2005

Components Start with integrated conceptual model—O3, PM, RH Source apportionment of STN and IMPROVE Organic speciation and source apportionment – urban and rural Ammonia/nitric acid – which precursors control PM? PM trends Episode analysis Transport and trajectory analysis Ozone weekend/weekday differences, trends, NOx/VOC limitation

Source apportionment of IMPROVE and STN data Work by Eugene Kim and Phil Hopke: PMF, ME, CPF, PSCF

Bondville, IL Bondville IMPROVE site Mar. 2001 – May 2003 9 sources Source profiles Bondville IMPROVE site Mar. 2001 – May 2003 9 sources

Averaged source contributions Bondville, IL Averaged source contributions

Time-series source contributions Backward trajectories Bondville, IL Time-series source contributions Backward trajectories 4/22/2001 7/1/2002

Weekday/weekend variations Bondville, IL CPF plots Weekday/weekend variations

8 carbon fraction profiles Gasoline emissions include the lower temperature carbon fractions (OC1 - OC4). Diesel emissions contain large amounts of the EC1 Gasoline vehicle emissions Diesel emissions

Secondary sulfate aerosol: Bondville Summer-high: Tennessee, Ohio river valley, Ohio, Louisiana, Alabama Carbon-high: Canadian forest fire, Nebraska, Dakota, Minnesota, Iowa Winter-high: Nebraska, Kansas, Oklahoma, Texas, Tennessee valley, Ohio river valley

Inter-site comparison General summary of the source contributions Secondary sulfate The largest contribution to the PM2.5 Secondary nitrate, Motor vehicle The 2nd largest contribution to the PM2.5

Gasoline vehicle emissions No strong weekday/weekend variation Rural sites show lower conc.

LADCO Organic Speciation Projects 1. Seney National Wildlife Refuge Analysis of existing speciation carbon channel filters (monthly composites) for 2002 Collection and analysis of 28 high volume filters for summer 2002 2. Urban organics Collection and analysis of 1/6 day, high volume samples at Indianapolis, Detroit, Cincinnati, Chicago, St. Louis, and Bondville, 2004 calendar year

Primary emission tracers

Urban organic speciation sites Detroit, MI Northbrook, IL Cincinnati, OH Bondville, IL Indianapolis, IN St. Louis, MO

Data Overview At six sites (five urban, one rural) during Dec 2003 – Feb 2005, collocated with STN measurements, quartz fiber filters (QFFs) sampled QFFs analyzed for OC/EC Monthly composites of QFFs for speciation done at all sites Indianapolis selected for organic speciation of all 60+ samples during period At Indianapolis and Bondville, 3-hr DRUM data taken during sampling days

Seasonal Composition - Indianapolis Median OMC>SO4 NO3>OMC>SO4 OC graphed as OMC=1.4*OC

Weekday-weekend: OC OC has no significant trend Northbrook – Feb 04 – Jan 05 Cincinnati – Feb 04 – Jan 05

Weekday-weekend: EC EC shows clear weekday-weekend trend Detroit – Feb 04 – Jan 05 Cincinnati – Feb 04 – Jan 05

Urban OC Summary A large, generally high-quality data set is being assembled PM2.5 concentrations were sufficiently high and variable during the sampling campaign to be of interest to policy makers While STN, IMPROVE and other data routinely collected are useful, additional OCEC and speciated organics data will greatly help source apportionment efforts

Midwest Ammonia Monitoring Despite importance of ammonia in atmospheric chemistry, no national studies or routine monitoring of ambient (non-source-influenced) concentrations has been done To fill data gap, MRPO and CENRAP began rural monitoring in Oct. 2003 Beginning network--10 sites: 9 rural, 1 urban Current network – 12 sites, 8 rural, 3 urban All rural sites are IMPROVE sites except Pleasant Green, MO Denuder/filterpack sampling Phosphoric acid coated denuder for NH3 Sodium carbonate coated denuder for HNO3 and SO2 Teflon filter followed by nylon filter (for nitrate dissociation) Two sites (Pleasant Green MO, Lake Sugema IA) use automated R&P samplers (different denuder/inlet configurations) 1/6 day sampling Two continuous samplers, Pranalytica and IC, at Bondville for QA Both NH3 and HNO3 very difficult to measure, due to reactivity, ‘stickiness’, low ambient concentrations, significant inlet losses Too soon to compare with IMPROVE – no data for IMPROVE posted yet for this period Goals: Characterize rural ambient NH3 (and related parameters) because very few data exist. Need spatial variability, temporal variability (seasonal, diurnal), and relationship of NH3, HNO3 with other gases and PM species.

LADCO-CENRAP NH3 Network Red=urban sites Blue=rural sites

Temporal variation in NH3 Log scale Base F2 NH3 Emissions

Spatial variation in NH3

Predicted PM2.5 Concentrations (Blanchard & Tanenbaum) At current NH3, PM is about equally sensitive to reductions in SO4 and NO3 At current SO4 level, PM is slightly more sensitive to NH3 reductions than to HNO3 (at this site; sites further west are more sensitive to HNO3 reductions); 30% reduction in NO3 leads to .75 ug less pm, 30% reduction in NH3 leads to 1 ug less PM PM is about equally sensitive to reductions in SO4 and NO3 PM is slightly more sensitive to NH3 reductions than to HNO3

Sample-by-sample evaluation of gas ratio and excess ammonia Nitric acid limited Nitric acid limited Sample-by-sample evaluation of gas ratio and excess ammonia (Blanchard & Tanenbaum) Nitric acid limited Nitric acid limited

Preliminary Conclusions At current conditions, PM mass decreases in response to reductions in sulfate, nitric acid, and ammonia At current conditions, particulate nitrate formation limited generally by nitric acid At expected future conditions (i.e., lower sulfate), PM mass is more responsive to nitric acid reductions Daily and hourly data support same conclusions

Future Applications for NH3 Data Model evaluation (Kirk Baker dissertation) Inventory validation – daily variability, seasonal variability, diurnal variability Correlations with meteorology, other AQ parameters PM episode analysis – Feb. ’05 event, role of snow as source/sink

PM Trends

Seasonal Urban Trends

Weekday/Weekend Ozone Differences (Blanchard&Tanenbaum)

Trajectories, Transport

Episode Analysis Based on AIRNOW From Rudy Husar Time pattern of the 0502 Event The overall event lasted about 10 days, Jan 28-February 7 The Upper Midwest peaked first (Jan 31-Feb 2); Industrial MW later (Feb 3 – 6) The Industrial MW region show more diurnal variation (lowest in the mid-afternoon) 0502 PM Event PM Event

24 hr Average PM-2.5 Concentrations in Michigan (1/26/05 thru 2/10/05) warm air advection cold front approaches From Neal Conatser, Mich. DEQ strong cold front passes weak cold front passes