Posting type Advisory
Subject Carbon analyzer signal integration threshold modified
Module/Filter URG 3000 / Quartz 25mm
AQS Parameter 88312, 88320, 88321, 88324, 88325, 88326, 88327, 88328, 88329,
Codes 88330, 88331, 88355, 88357, 88370, 88374, 88375, 88376, 88377,
88378, 88379, 88380, 88381, 88382, 88383, 88384, 88385, 88388
Sites Entire CSN network
Period January 1, 2016 through September 12, 2017 sample dates
Recommendation Information only
Submitter N. Hyslop, nmhyslop@ucdavis.edu
Starting with January 2016 sample dates, Desert Research Institute (DRI) switched from Model
2001 to Model 2015 carbon analyzers for the quartz filter samples. The new Model 2015
analyzers measure carbon with a Nondispersive Infrared [NDIR] carbon dioxide (CO2) detector,
where the old Model 2001 analyzers had used Flame Ionization Detection (FID) of methane
(CH4). After examination of several months' worth of data, DRI determined that the Model
2015 carbon signal integration threshold differed from that of the Model 2001. This difference
mostly affected low elemental carbon (EC) concentrations. Further data analysis and testing by
DRI determined that a revised Model 2015 threshold for fractions organic carbon 1 (OC1)
through EC2), with an unchanged threshold for EC3, was more equivalent to the carbon
detection sensitivity used in the Model 2001 analyzers. The reprocessed 2016 average EC
concentration increased by 0.03 |ig/m3, which is similar to the lower quantifiable limit based on
the field blanks (0.024 |ig/m3). The threshold change has a minor effect on average carbon
levels. Figure 1 shows the original versus reprocessed loadings for EC and OC; the differences
are not visible on a linear scale. Figure 2 shows the same data on a log scale, illustrating there
are differences at the lowest concentrations, particularly below 10 |j,g/filter (which corresponds
to 0.3 |j,g/m3 assuming a nominal sample volume of 32 m3). Figure 3 provides another way of
looking at the differences; the concentration data were binned into 20 groups (5th percentiles), the
bottom plot shows the average concentration for each bin, and the top plot shows the average
relative difference between the original and reprocessed carbon loadings for each bin.
After reviewing the differences in the carbon concentrations with the old and new thresholds and
the minor impact on the data, the decision was made to not reprocess or redeliver the data to
AQS. The differences are too small to warrant the effort and disruption that would ensue if the
values were changed. The new integration threshold was implemented beginning with samples
collected on September 13, 2017. This advisory serves to inform data users of the change in the
integration threshold and slight bias in January 1 - September 12, 2017 measured values toward
under-reporting at low carbon concentrations.
1
-------
SampleType Field Blank ° Sample
Figure 1. Plots comparing the Original (y-axis) and Reprocessed (x-axis) carbon filter loadings
for elemental carbon (ECTR) and organic carbon (OCTR) on linear scales.
SampleType Field Blank ° Sample
Figure 2. Plots comparing the Original (y-axis) and Reprocessed (x-axis) carbon filter loadings
for elemental carbon (ECTR) and organic carbon (OCTR) on log scales.
2
-------
— -20-
<1>
0
c
0)
1
b
CD
>
TO
-40"
-60-
• I
..!:
::::!
• t
ECTR
OCTR
TC
s
. . . ! :
*
. • • •
•
•
—i 1 1 ¦ 1 ¦ 1 1 1
1
1—
—i
i_
ci)
CD
100-
•W
D>
C
~o
CO
o
CO
10-
CO
03
c
03
CD
>
1-
0s"
LO
O
o
in
*->¦
o
O
CM
o^-
LT)
CM
O
CO
in
CO
o
^r
LO
¦^r
o
LO
<3**
LO
LO
o
CD
0s
m
CO
o
LO
r--
a
--p
0s
m
oo
o
o>
m
o>
m o m o
t- CN (M CO
o m o m o m
rr in in to
------- |