% IMPORTANT: The following is UTF-8 encoded. This means that in the presence
% of non-ASCII characters, it will not work with BibTeX 0.99 or older.
% Instead, you should use an up-to-date BibTeX implementation like “bibtex8” or
% “biber”.
@PHDTHESIS{ToengesSchuller:41864,
author = {Toenges-Schuller, Nicola},
title = {{G}lobale {V}erteilungsmuster anthropogener
{S}tickoxidemissionen: {V}ergleich und {I}ntegration von
troposphärischen {S}atellitenbeobachtungen und
{M}odellrechnungen},
volume = {4160},
issn = {0944-2952},
school = {Univ. Köln},
type = {Dr. (Univ.)},
address = {Jülich},
publisher = {Forschungszentrum Jülich GmbH Zentralbibliothek, Verlag},
reportid = {PreJuSER-41864, Juel-4160},
series = {Berichte des Forschungszentrums Jülich},
pages = {VIII, 140 S.},
year = {2005},
note = {Record converted from VDB: 12.11.2012; Köln, Univ., Diss.,
2004},
abstract = {Nitrogen oxide (NO$_{x}$, NO+NO$_{2}$) plays a key role in
tropospheric chemistry, for example as an important
precursor of ozone. Its distribution is studied with global
chemistry transport models, which need surface emission data
as input. A widely-used emission inventory is the EDGAR
database, where the anthropogenic emissions are estimated
using economic data of the individual countries. This
database contains large and mostly unknown uncertainties. In
this thesis, satellite data are used in addition to EDGAR to
estimate the geographical distribution pattern of
anthropogenic NO$_{x}$ emissions. Due to the short
tropospheric lifetime of NO$_{x}$ (≈ 1 day), its global
distribution is highly correlated to the distribution of the
emissions, which allows using measurements of tropospheric
NO$_{2}$ column densities by the satellite instrument GOME
as a proxy for anthropogenic emissions in the areas
dominated by those emissions. Two GOME evaluations are used
in this thesis, one done by Richter \& Burrows (IUP Bremen)
and one done by Leue et al. (IUP Heidelberg). As yet another
proxy for anthropogenic emissions, calibrated satellite
measurements of the nighttime lights of the world
(Operational Linescan System (OLS), Defense Meteorological
Satellite Program, NGDC) are used. Within this thesis, a
method is developed to calculate pattern errors using the
correlation coefficients of at least three fields with
independent errors (correlation error analysis). The pattern
error of a field is defined here as the ratio of the
variance of the error contained in that field to the
variance of the total field. At first, the correlation error
analysis is applied to the annual mean values of four
NO$_{2}$ column density fields in those areas dominated by
anthropogenic emissions: Two GOME evaluations and two model
calculations done with the global chemistry transport model
MOZART. The first model calculation was done using the
model’s standard emission fields which are based on the
EDGAR database; in the second calculation, the anthropogenic
NOx emissions were replaced by a source based on the
satellite images of the nighttime lights of the world. Since
neither the errors of the two GOME evaluations (same
instrument, similar evaluaiii tion algorithms) nor the
errors of the two model calculations (done by the same
model) are independent, only error ranges can be given for
the column density fields: The pattern errors of the two
model calculations range from 18\% to 50\%, the pattern
error of the GOME evaluation by Richter \& Burrows ranges
from 0\% to 39\% and the pattern error of the GOME
evaluation by Leue et al ranges from 26\% to 55\%. For the
correlation error analysis of the emission fields, there are
three independent sources available: EDGAR, OLS and GOME. To
at least partly eliminate the effect of undirected
transport, the GOME fields are deconvoluted and risen to a
higher power. This sharpens the patterns of the satellite
measurements when interpreted as emissions. If outliers in
the source fields are eliminated before applying the
correlation error analysis, the pattern errors of the four
fields determined in this thesis read as follows: EDGAR
anthropogenic: (27±5)\%, light-based NO$_{x}$ source: (26
± 5)\%, NO$_{x}$ source GOME Richter: (33 ± 5)\%, NO$_{x}$
source GOME Leue: (45 ± 5)\%. So far, the error estimates
for EDGAR were rather rough; the error specifications for
the GOME fields can help to improve the retrieval
algorithms. Finally, the four emission fields are combined
minimizing the pattern error of the combination field.
Assuming that the pattern errors that were determined in the
areas dominated by anthropogenic emissions are the same in
other regions as well, an anthropogenic NO$_{x}$ emission
field with global coverage can be constructed: In the areas
dominated by biogenic emissions, only the anthropogenic
EDGAR source and the light-based emissions are combined. In
the areas dominated by anthropogenic emissions, all four
fields are combined where possible. At grid points where one
or two fields show outliers or are undefined, only the other
fields are combined. The pattern error of the combination
field amounts to (15 ± 2)\%, which is a considerable
reduction compared to the pattern errors of the four
original fields. The combination field is unique only up to
a constant offset and a constant factor, since only the
pattern of that field is fixed by the construction. This
offset and factor are chosen relative to the anthropogenic
NO$_{x}$ emissions of EDGAR. With this source field a MOZART
model calculation is done. The spatial correlation of the
annual mean of the tropospheric column densities of this
field with either of the GOME evaluations is higher than
that of the original model calculations with either the
EDGAR or the light-based NO$_{x}$ emissions.},
cin = {ICG-II},
cid = {I:(DE-Juel1)VDB48},
pnm = {Chemie und Dynamik der Geo-Biosphäre},
pid = {G:(DE-Juel1)FUEK257},
typ = {PUB:(DE-HGF)11 / PUB:(DE-HGF)3},
url = {https://juser.fz-juelich.de/record/41864},
}