Measurements of CO Tropospheric Burden From the Ground and From a Satellite: Error Analysis

1
Measurements of CO Tropospheric Burden From the
Ground and From a Satellite Error Analysis
2004 JOINT ASSEMBLY of AGU, CGU, SEG, and EEGS,
Montreal, 20 May, 2004.

• L. N. Yurganov,
• Frontier Research System for Global Change,
  JAMSTEC, Yokohama, Japan
• A.V. Dzhola, E. I. Grechko,
• Obukhov Institute of Atmospheric Physics, Russian
  Academy of Science, Moscow, Russia,
• D. P. Edwards, J. C. Gille,
• Atmospheric Chemistry Division, National Center
  for Atmospheric Research, Boulder, CO, USA
  J.R.Drummond,
• University of Toronto, Toronto, Canada

2
Problem of interest
Total column amount of CO integrated over vast
areas (e.g., NH or High NH) is the atmospheric
(or tropospheric) burden. This burden actually is
a result of an equilibrium between sources and
sinks. A useful information about the
hemisphere-averaged parameters of CO budget may
be retrieved from the burden, especially from the
burden variations (deviations form the
equilibrium). We presented yesterday an analysis
of CO burden anomalies between 1996 and 2004.
How accurate are the burdens measured from space
and from the ground and what are the
corresponding errors in retrieved emission
anomalies?
The first class of errors is connected with
different altitude sensitivity of ground-based
spectrometers and MOPITT.
The second source of errors (mostly for the
ground-based instruments) is connected with a
sparse network of the observational sites.
3
Some results, presented yesterday, at the A33B
section.
CO measurements to the North from 30ºN Anomalies
of monthly means, averaged over 16 CMDL stations
in the BL, 5 FTIR (low altitude), 5 in-situ 2
FTIR in the mountains, compared to MOPITT, in
ppb (SCO / Sair)
Total burdens converted into emissions Top panel
anomalies of burden, Bottom panel anomalies of
emission (life-time is standard)
Forest fires emissions in HNH (Anomaly normal
normal 50.7 Tg/year)
4
MOPITT
First, lets look at vertical sensitivity of the
MOPITT and ground-based data.
Simultaneous measurements in Russia in period of
peat fires
Spectrometer
1.5 km
17
1.15 ppm in BL
Peat fires
Zvenigorod
60 km
Moscow
5
? Ground based spectrometers measure CO total
column amounts for the case of highly polluted
boundary layer with a sufficient accuracy (17
error for the worst case). The MOPITTs ability
for measuring boundary layer CO is questionable.
? However, locations of spectrometers are
randomly scattered over the globe. Sometimes they
are too close to source areas and their data are
not representative, sometime they are beyond them
and provide nothing about CO over burning areas.
Also there are no FTIRs in the oceans!
? Lets estimate what may be a contribution of CO
over burning areas for the HNH tropospheric CO
burden. The most uncertain is the area with the
BL impacted by the forest fires emissions. It
should be larger than areas burned, but, as we
saw for the case of Zvenigorod, even at the
distance of 100 km there were days with low
CO.
6
These MODIS and MOPITT data a) monthly mean CO
total column b) Fine mode aerosol optical
depth c) Fire counts are presented just to
illustrate the chosen magnitude of the impacted
area. Lets assume that the rectangle at the
bottom panel coincides with the region impacted.
Its area is 12º x 100º x cos (50º) x 111 x 111
9.4 millions km2. For a comparison area burned
in 1998 in the HNH was estimated as 0.16 millions
km2 (Sukhinin et al., 2004).
7
Extra CO over Zvenigorod due to the BL
pollution (4.2-2.6) E18 1.6 E18
mol/cm2
Extra CO total column in the BL multiplied by 9.4
million km2 and converted in the CO mass amounts
to 7 Tg.
8
Lets consider a possible error. Maximum area
with polluted BL is assumed in September. Extra
burden was calculated in assumption of 1.6 E18
mol/cm2 for the polluted area.
Anomalies
Red line is the corrected retrived emission
Assumed area with impacted BL
Red line is the burden corrected for BL pollution
The error due to underestimated burden
Maximum error in the anomaly of the emission rate
for June-August was around 10.
9
Annual emission anomalies based on entire MOPITT
measurements (green full rectangles) and just on
continental data (empty green rectangles)
expanded onto HNH.
Total column anomalies measured by MOPITT and
averaged over different regions.
10
CONCLUSIONS
?MOPITT data are in general accordance with other
available data. All of them reveal enhanced CO
abundance in 2002 and 2003.
? CO High Northern Hemisphere burden anomalies
directly measured by MOPITT are mostly close to
anomalies obtained by expanding mountain stations
data (FTIR and in-situ) onto the entire
semi-hemisphere. MOPITT mean burden anomaly for
2002-2003 is 40 less then that obtained from low
altitude spectrometers (10 Tg vs 14 Tg).
Emission anomalies in 2002 and 2003 retrieved
from MOPITT are 30-40 lower than those from
FTIR. A part of this difference is due to lower
total columns over oceans (MOPITT averaged over
continental areas is just 15-20 less than FTIR).
? A comparison MOPITT vs spectrometer in a
source area (near burning peat lands whith mixing
ratio in the boundary layer up to 1 ppm or
higher) shows a significant underestimation of
MOPITT data and a smaller overestimation of
spectrometer data (less that 17) when the same
standard a priori profiles are used.
11
? An estimate of a possible contribution from the
BL areas polluted by emissions from biomass
burning strongly depend on the magnitudes of
impacted areas that are absolutely unknown. As an
example, 10 millions square kilometers and extra
columns measured in Zvenigorod, lead to 10-error
in estimated hemispheric emission anomaly.
? For the future we recommend to keep the
existing network of FTIRs, even to expand it.
Satellite measurements, being quite satisfactory
even in present form, has an advantage of full
surface coverage. Ground based measurements are
more accurate and can serve as a standard
validation tool for the future space based
instruments.