Overview of CSN Data Relevant to OCEC Artifact Adjustments presented by James Flanagan RTI Internati

1
Overview of CSN Data Relevant to OC/EC Artifact
Adjustmentspresented byJames FlanaganRTI
InternationalDavis, CAJanuary 22-23, 2008
2
Assignment Review of current artifact
adjustment approach for CSN, including
sampler-specific variations

• There is no artifact adjusted data currently
  being posted to AQS for any CSN species,
  therefore this presentation will focus on
• Blank data available that could be used for
  artifact adjustments
• Original CSN samplers and analysis
• New 3000N samplers and analysis
• CSN program timeline
• CSN sampler types
• Blank level differences between sampler types
• Lab blanks and other QC data

3
CSN Timeline

• 2/9/2000 7/7/2003
• OC, EC analyses done by RTI International
  Laboratory using Sunset Labs Instruments
• No OC or EC fractions (peaks) reported
• OCX or OCX2 were reported during this period
• 7/8/2003 present
• OC, EC analyses for CSN samplers done by RTI
  International Laboratory.
• 4 OC fractions (plus Pyrolysis C) are reported
  no OCX or OCX2.
• no EC fractions reported.
• -- Initial Set of 55 3000N Samplers Added to CSN
  in May, 2007 --
• 5/2007 present
• 3000N filters are being analyzed by Desert
  Research Inst. (DRI).
• 3000N samplers are collocated with the existing
  CSN samplers at each site, and are run collocated
  for 2 months
• 4 OC, Pyrolysis C, and 3 EC fractions, plus OC
  and EC by TOT and TOR are being uploaded to AQS
  without artifact correction.

4
CSN STN TOT Analysis Method

• Filter 47mm Whatman quartz w/ 5 borosilicate
  binder
• Filters pre-fired at 900oC in air
• Sunset Lab Carbon Aerosol Analyzers
• Protocol NIOSH 5040, optimized for PM2.5
• Time-driven analysis (12 min total)
• Thermal-Optical Transmittance (TOT)

5
CSN Sampler Types

• URG MASS Wins Impactor Teflon module
• Andersen RAAS - cyclone Teflon coated module
• MetOne SASS mini-cyclone alloy module
• other types
• RP 2300 internal greased impactor alloy
  module
• RP sequential FRM

6
CSN Sampler Module Types
URG MASS
Andersen RAAS
Cassette used for IMPROVE and 3000N Samplers
MetOne SASS
RP 2300
7
CSN Field and Trip Blanks(reported to AQS)

• Field Blanks are collected at 10 of routine
  frequency (monthly or bimonthly at each site)
  filter modules mounted for a short period of time
  and removed.
• Trip Blanks are collected at 3 of rout. freq.
  filter modules not removed from shipping bag.
• Field and Trip Blank results are very similar.
• Field/Trip Blank levels differ significantly
  between sampler types.

8
CSN Field/Trip Blank Averages(Original OC/EC
Analyses)
Trip and Field Blank averages 2000 2006,
ug/filter

RP module contains silicone impactor grease,
which may appear as EC.

• 47 mm Whatman filter with borosilicate binder.
• CSN (STN) analysis protocol.
• Each number is 100s or 1000s of measurements.

Source Max R. Peterson, James B. Flanagan, Larry
C. Michael, and R.K.M. Jayanty. 2007. Analysis
of PM2.5 Speciation Network Carbon Blank Data.
AAAR 26th Annual Conference, Reno, NV, September
24-28.
9
CSN Laboratory QC Data for Original OC/EC
Analyses(not reported to AQS)

• Analyses are performed by RTI Internationals
  OC/EC Laboratory in North Carolina using Sunset
  Labs Instruments.
• Acceptance Blanks full STN analysis of the
  larger of 2 or 2 of pre-fired filters.
• Instrument Blanks full STN analysis using a
  clean filter punch, usually from the preceding
  analysis.
• Duplicate Analyses 10 of filter samples
  reanalyzed on the same analyzer to test
  uniformity of filter deposit.
• Replicate Analyses a group of filters analyzed
  on 3 or more analyzers to assess between-analyzer
  variability and to establish measurement
  uncertainties.

10
CSN Laboratory QC Data Received for OC/EC
3000N(not reported to AQS)

• Analyses are being done by Desert Research
  Institute (DRI) in Reno, NV.
• System Blanks Full IMPROVE_A analysis without a
  filter punch.
• Replicate Analyses Analysis of a second punch
  from 10 of filter samples on a randomly-chosen
  (different or same) analyzer used to evaluate
  between-analyzer variability.
• DRI performs additional QC not reported to RTI.

11
CSN Available Blank Types with 3000N Sampler
Protocol(blank data reported to AQS)

• Afterfilters identical to IMPROVE NEW to CSN
• IMPROVE-type field blanks (passive blank in 4th
  position of cassette) NEW to CSN
• Field Blanks (like existing CSN Trip and Field
  Blanks)

12
Afterfilters Frequency and Availability of Data
for Statistics

• Since May, 2007, CSN runs afterfilters at all
  sites equipped with 3000N at a rate of 10
• Sufficient afterfilter data are available to
  calculate reliable statistics on a monthly basis
• Median
• Standard Deviation
• other?

13
TOT Afterfilter vs. Ambient Levelswith 3000N
Module

• Original CSN field and trip blanks are generally
  poorly correlated to season or site location
  however, afterfilters from 3000N show some
  correlation with ambient levels.

• Data from 3000N
• X-axis front filter
• Y-axis afterfilter
• N 820, R20.2378
• 55 sites nationwide
• 5/2007 12/2007

14
CSN Temporal Variability in Trip and Field
Blank Levels
Average Total Carbon Blank by Year, ug/m3 (subset
of sites with long-term data)

• Between 2002 and 2005, STN field and trip blank
  annual averages for Met One (SASS) and Anderson
  (RAAS) samplers decreased 33 and averages for
  URG (MASS) samplers increased 25.

From Carbon Measurements and Adjustments ,
presented by Neil Frank at Air Quality Data in
Health Effects Research conference, Nov 30 Dec
1, 2006
15
From Carbon Measurements and Adjustments ,
presented by Neil Frank at Air Quality Data in
Health Effects Research conference, Nov 30 Dec
1, 2006
16
CSN Temporal Variability in Trip and Field
Blank Levels
MetOne Sampler
Total Carbon Field and Trip Blanks, µg/m3
From Carbon Measurements and Adjustments ,
presented by Neil Frank at Air Quality Data in
Health Effects Research conference, Nov 30 Dec
1, 2006
17
Lab Blank Variability
Second
Third

• Instrument blanks are typically less than 10 of
  trip and field blanks when expressed in the same
  units
• Between-instrument variability and the effect of
  a software upgrade are visible
• One of the lab analyzers appears to be more
  sensitive to the software change than the other.

Source Max R. Peterson, James B. Flanagan, Larry
C. Michael, and R.K.M. Jayanty. 2007. Analysis
of PM2.5 Speciation Network Carbon Blank Data.
AAAR 26th Annual Conference, Reno, NV, September
24-28.
18
Notable Differences Between Original CSN Method
and IMPROVE/3000N

• Face velocity and deposit density are potentially
  up to 10x greater for IMPROVE/3000N relative to
  MetOne due to flow rate and filter size.
• Different filter materials.
• Different flow control technology (IMPROVE)
• Different analysis protocols
• temperature steps
• timebase
• TOT vs. TOR

19
Summary Slide Pros and Cons for the use of
existing field and trip blank data for artifact
correction of original CSN data

• Pro Lots of existing data 2000 - present
• Pro Probably closer to truth than lab blanks
• Con May not represent true artifact
• Con Questionable relevance to IMPROVE data
• Con Unlike afterfilters, the existing CSN trip
  and field blanks are not sensitive to seasonal or
  geographic variations