Data Quality Issues with Continuous PM2'5 and AIRNow

1
Data Quality Issues with Continuous PM2.5 and
AIRNow

• Lewis Weinstock
• U.S. EPA/Office of Air Quality Planning and
  Standards
• Research Triangle Park, North Carolina
• weinstock.lewis_at_epa.gov

2
PM2.5 Mapping is Growing ...
305 Active U.S. Monitors
3
rapidly
425 Active U.S. Monitors
4
and so is the diversity of methodsreporting to
AIRNow
Monitors approved for public reporting (1/25/05)
5
Geographic Distribution of AIRNow PM2.5 Methods
Beta Attenuation
Missing
Surrogate Measure
TEOM
Monitors approved for public reporting (1/25/05)
TEOM-FDMS
6
FRM-like The process of statistically transforming
continuous PM2.5 data to match the
FRM. EPA-454/B-02-002 R2gt0.81 Overall Method
Bias lt10
7
The FRM-like Transformation

• Sounds easy but may be challenging.
• Many organizations (and vendors) have developed
  creative approaches to adjust for
• Instrument effects like sample heating.
• Seasonality, especially when cold.
• Geographic effects on aerosol content.
• NAQC presentations by Puget Sound Clean Air
  Agency and Missouri DNR.
• NESCAUM summary by George Allen.
• How often should relationships be reviewed and
  updated?

Courtesy Henry D. Felton (Dirk) NYS DEC, Albany
NY
8
11
9
(No Transcript)
10
11
11
(No Transcript)
12
11
13
(No Transcript)
14
How often should FRM-like relationships be
updated?
15
Other Data Quality Issues with Continuous PM2.5
16
Case Study Following the Errant Red Ball
17
Valid or Suspect?
Analysis Site operators had difficulty replacing
TEOM filter on 1/26/05 1/27/05. Invalid data
stream was caught by DMC on 1/29/05 and site
disabled in AIRNow. Problem resolved in field
1/31/05.
18
Flagging at the Site or AIRNow (DMC) Level
Diagnostics Buddy Sites
1/25
1/26
1/27
1/28
1/29
1/30
1/31
2/1
19
When Good Operators Go Bad
Reseat TEOM filters
Change PM2.5 And PM10 Filters
20
Rapid Filter Overload Syndrome in NC

• TEOMÒ filters overload (pressure drop from 20
  to 90) in 1-3 days instead of typical 10-20
  days.
• Occurs during periods of high ambient PM2.5.
• At least six overload episodes occurred in NC
  during summer of 2004.
• Multiple sites often affected at same time.
• Leads to overloaded filters, flow restrictions,
  and invalid data collection.

A
B
C
A
Unused filter
Exposed filter, routine change after 14 days,
load 21 50, mass666 mg
B
Exposed filter, lasted 5 days, load 18 112
(overloaded), mass531 mg.
C
21
PM Data Behavior During Rapid Filter Loading
PM2.5gtPM10
DFilter
PM2.5 Filter Load Progression
39 40 41 42 43 45 48 52 57 63 69 75 81 86 91 95
99 100
22
Take Away Message

• Agencies are successfully reporting FRM-like data
  to AIRNow, but
• Periodically review FRM-continuous relationships.
• Assess new monitoring technology and network
  needs.
• Keep information table current in AIRNowTech (add
  the capability to track changes in FRM-like
  equations?).
• Review maintenance and real-time validation
  procedures with site operators.
• Use digital technology to remotely monitor
  diagnostics.
• Take better advantage of automated internal
  (pre-DMC) and external quality assurance routines
  for filtering real-time PM2.5 data.
• Customize AIRNowTech QC criteria like Buddy
  Sites, rate-of-change, sticking values, and max
  suspect/severe.