Application of Combined Mathematical and Meteorological Receptor Models (UNMIX

1
Application of Combined Mathematical and
Meteorological Receptor Models (UNMIX Residence
Time Analysis) to 1991-99 IMPROVE Aerosol Data
from Brigantine NWR, NJ (R.Poirot P. Wishinski,
VT DEC) (also see upcoming Comparative Positive
Matrix Factorization Analysis by
Jong Hoon Lee Barbara Turpin at Rutgers
University)
The general Approach involves applying
Mathmatical Model to Speciated Aerosol Data, then
applying Ensemble Trajectory Techniques to
evaluate sources (See Sources of Fine Particle
Concentration and Composition in Northern
Vermont, posted at http//capita.wustl.edu/nearda
t/reports/ TechnicalReports/ReceptorModels/
awmatoxRP.pdf)
2

• Traditional UNMIX Approach for Apportioning
  Measured Fine Mass (employed at Underhill, VT)
• Include Measured Mass as Model Input and
• Specify It as the Total and Normalization
  Variable
• Compositions Contributions expressed as Mass
  Fractions
• Non-Traditional Approach for Apportioning
  Reconstructed Mass Species (employed at
  Brigantine, NJ)
• Exclude Measured Mass as Model Input and
• Construct Composite Variables for Sulfates,
  Nitrates, Carbonaceous Matter (EC 1.4 OC) and
  Crustal Material
• 4 Separate UNMIX Runs, 1 for each Major Mass
  Species
• Initial Results expressed as fractions of Major
  Species

3
Brigantine 4 Separate UNMIX Runs, 1 Each to
Explain Mass of Sulfates (1.375xSO4),
Carbonaceous Matter (EC1.4xOC)
Nitrates (1.29 x NO3) Crustal
Material (4 x Si) 22 Sources, expressed as mass
fractions of 4 major species
4
Half the Initially identified 22 sources were
highly correlated (same source was
identified as contributor to several major
species) Combining these Redundant Sources
reduced from 22 to 11
5
(No Transcript)
6
Redundant Sources had highly correlated daily
mass contributions and also had similar
compositions for common trace elements
7
Sum of daily source contributions accounts for
90 of measured mass and perfectly explains
100 of the reconstructed mass
8
Average Monthly Source Contributions at
Brigantine NJ, 1991-99
9
Fine Mass Fractional Source Contributions at
Brigantine NJ, 1991-99 on Worst
20 (20 ug/m3) Average (11 ug/m3) and Best 20
(5 ug/m3) PM-2.5 Days
10
Multi-year Changes in Source fine Mass
Contributions (Note most of the improvement
results from pre-1996 reductions in Summer
Sulfate)
11
Thermally Stratified Carbon Fractions Show
Different Distributions
12
Four of the sources show day of week
differences Zn Oil gt on Wednesdays, Woodsmoke
Lite Carbon are gt on Saturdays. Of these
4 Sources Zinc gt Wed all year round, Oil gt Wed
only in Winter, Woodsmoke gt Sat only in
Winter Lite Carb gt Sat only in Summer
13
Wind Roses based on hourly surface met from
Atlantic City, for every sample day, and then
constrained to days when each source was high
(top 10 days). Met data Rose Works software
kindly provided by Steve Mauch, at UAI
Environmental.
14
(No Transcript)
15
(No Transcript)
16
(No Transcript)
17
Based on ATAD Trajectories kindly provided by
Kristie Gebhart, NPS
18
Time Series of Soil Dust Impacts at Brigantine
Occasional Soil Dust Spikes hit Brigantine a
few times every Summer (July) and Always
correspond with (much) Higher soil spikes to the
Southeast...
19
(No Transcript)
20
No relationship between AlCa Ratio and Fine Soil
Mass at Brigantine, until Al/Ca gt 3.8. An Al/Ca
Ratio of gt 3.8 and a fine soil concentration of gt
3 ug/m3 have been identified as Indicator of
Sahara Dust. For Example in Gebhart, Kreideneweis
Malm analysis of Big Bend, TX aerosol,
currently in press in The Science of the Total
Environment. (draft, do not
cite, etc.)
21
Sea Salt (normalized to Na) derived in 7 Separate
UNMIX Runs, with 7 Different Subsets of Input
Variables and Observations, with Nearly
Identical Compositions (for common Input
Variables) and Nearly Identical Contributions
(for common Input Observations)
(Chloride Not Included because its mass
fractions Too Variable)
22
The 5 Extra Sea Salt Runs also Yielded 5 extra
Soil sources. These agree almost Perfectly
with the original Soil source derived from
Crustal (Si) Run, Both in terms of their
daily mass contributions and fractional
mass compositions, for common observations and
common input variables. So these Soil Sea Salt
sources may be wrong, but they are robustly
identified in multiple runs with multiple combos
of input variables and observations.
23
Sea Salt Windblown Dust are both contaminated
by NO3 SO4 (How would we calculate
Extinction Efficiencies for these Mixtures?)
and who do we Blame for the associated
Visibility effects?