Climate Change Impacts on Terrestrial and Aquatic Ecosystems.pptx
ICOS ATC Labeling Process
1. ICOS ATC LABELING PROCESS
C. Yver Kwok, O. Laurent, C. Philippon, K. Yala, L. Hazan, A. Amara, M. Ramonet , M. Lindauer, J.
Levula, M. Steinbacher, S. Conil, G. Vitkova, O. Hermansen, M. Mölder, M. Heliasz, M. Ottoson
Lövfenius and L. Rivier
3rd ICOS science conference – 11-13th Sep 2018
2. LABELING PROCESS
• Labeling purpose
• gather data to optimize the station performances
• train the PIs to keep the quality high
• ensure that the required quality level is reached
provide harmonized and high precision scientific data on carbon
cycle and greenhouse gas budget and perturbations
5. • Check that the site follows the specifications for the initial test period (frequency of
calibration, frequency of target gas measurements,…)
• Regular data check together with the PIs through visioconference (Minutes available shortly
after)
• Exchange of emails
• Production of a report that looks at
• Regularity of data QC
• Data sampling distribution between air, target, flushing periods
• Stabilization for the calibration and target gases
• Calibration: drift, stabilization time between cycles
• Temperature dependence
• Availability of meteorological and diagnostic parameters
• Ambient air data (availability, trend, windroses, …)
• Uncertainties for the station and compared to other ICOS stations
• First inlet line integrity and water correction assessment tests
• Problems encountered and solved
INITIAL TEST PERIOD
6. • Goal: Evaluate the time needed for the instrument to flush the previous sample and
be stable for the current one. We look at the short and long term target and
calibration gases over the last 3-6 months. This allows to optimize the sampling
strategy and can also help pinpoint problems on specific parts of the piping system.
STABILIZATION TIME
7. STABILIZATION TIME
First statistics on the 11 ICOS sites from the STEP2 reports
CO2 CH4 CO
Usually stable faster as drier
Dry but only one injection every 15 days
Cal outlier
still present in
current data
LTT outlier fixed
since then
Due to 2 very noisy
instruments, currently
improved
8. • Evaluate the quality of the calibrations by checking the mean standard deviation
and difference between the different cycles of all the calibrations run during the
ICOS labelling step 2. Identify a calibration cylinder that would behave differently
from the others or one calibration sequence that needs further inspection and
quality control.
CALIBRATIONS
9. • During the initial test period, it is asked to run:
• 4 cycles of the calibration set (first cycle rejected for stabilization)
• Every 15 days
• 30 minutes each cylinder
• From the 11 station data, it was recommended to go on with:
• 3 cycles of the calibration set (first cycle rejected for stabilization)
• Every 15 days
• 30 minutes each cylinder for most stations
CALIBRATIONS
10. • Assess the presence or absence of biases and leaks in the station sampling system
• By injecting test gas at different locations of the sampling lines and checking the values
at these different locations
• By emptying the sampling lines from the top to the shelter and checking if the reached
vacuum stays stable
• Assess the time needed to flush the part of the lines that are not flushed at a high
rate by the flushing pumps
• Comparable to the stabilization time evaluated with the cylinders
• Assess the stabilization time for the whole unflushed line not just from the valve to the
instrument
INLET LINE INTEGRITY
11. • 10 stations labeled or in step 2 have performed this test
• Test for 8 sites showed no leaks or biases
• On 2 sites, tests allowed to highlight problems:
• a faulty tubing lead to a negative bias in CO2 for one level
• a water trap lead to a negative bias in CO2 for one level
• After changing the faulty parts, the test showed no more biases or leaks
INLET LINE INTEGRITY
12. Very high Target variability due non stable water
content --> this site used a nafion drier
After exchange with ATC, PIs changed the set-up in
May 2017 and got the water level more stable but no
change observed on the TGT variability
Then in December 2017, they removed as advised the
nafion altogether and solved the problem
SOLVING PROBLEMS
13. • During dataday, we observed a non-
expected gradient with 30m significantly
higher than the others
• PI identified the problem as coming from
the sampling intake that was in PTFE
• After replacing the part for all intakes, the
problem was solved
SOLVING PROBLEMS
14. • The labeling process allows
• ATC to know the PIs and their station better
• To make sure the protocols are followed
• To optimize the sampling strategy for each site
• To help detect and solve problems with the Pis
• To gather statistcial data to detect outliers
• To develop and improve the labeling process and the quality of the data
CONCLUSION