Before October 1, 1998, all
reported readings of CH4_LEL were obtained with a Drager
Multi-Pak. This was replaced with an
Industrial Scientific ATX-620 for all readings after October 31, 1998.
For readings of combustible
gas, the Multi-Pak has a stated
accuracy of +/- 4% LEL for readings below 40% LEL, and +/- 10% of reading for
readings between 40% and 100% LEL.
Readings above 200% LEL are out of range for the instrument and
considered highly suspect. On the ATX-620, readings above one million PPM (more
than 100% methane by volume) indicate the presence of ethane as well as
methane.
The ATX-620 has a stated
accuracy of +/- 20%, or one count if this is greater, over the range 0-100%
LEL, and +/- 15%, or two counts if this is greater, over the range 500 to one
million PPM. These accuracy figures
apply to temperature variation between -15º C. and +40º C. At the calibration temperature of 20º C.,
accuracy for both the catalytic and infrared sensors is plus or minus 5%. Ground observers estimate that most readings
were taken between -0º C. and 25º C, so that accuracy is somewhere between five
and twenty percent. This information
was not used in computing the statistics presented here, but may be useful to
readers in evaluating the quality of the data.
During the month of October
1998, readings were taken with both instruments as a calibration check. The readings were identical in 43 of 71
cases. In all but one of the remaining
27 cases, the readings from the ATX-620 were higher. Over all 71 cases, the reading on the ATX-620 averaged about 13%
higher than the reading on the Multi-Pak, with a standard error of about 17
percentage points. The discrepancy is
not larger than could be expected by chance, but to be conservative, all
methane LEL readings before November 1, 1998 were adjusted upward by ten
percent.
The maximum published range
for the Drager Multi-Pak is 100% of LEL.
Reported measurements of greater than 200% were considered by field
observers to be likely caused by sensor contamination and not a genuine
reflection of soil vapor composition.
Statistics were computed based on the measurements as reported, even
though the real values might have been much higher. Almost exclusively, this affected measurements in the period
before the cutoff date. It is
possible, therefore, that at some of the sites with an apparent increase, the
phenomenon might be an artifact caused by limitations of the measuring
device. A refinement for a future
version of this report will be to scan for and identify those sites with LEL
readings greater than 100% before the cutoff date. It is our general impression, however, that only a small number
of sites will be affected.