% 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”.
@MASTERSTHESIS{Rssler:256610,
author = {Rössler, Thomas},
title = {{O}ptimization and validation of atmospheric advective and
diffusive transport simulations},
school = {Fachhochschule Aachen, Campus Jülich},
type = {BA},
reportid = {FZJ-2015-06477},
pages = {II, 65},
year = {2015},
note = {Fachhochschule Aachen, Campus Jülich, Bachelorarbeit,
2015},
abstract = {Lagrangian particle dispersion models are indispensable
tools to study atmospheric transport processesbased on the
flow of individual air parcels.In operational uses cases
they are used to simulate the spread of radionuclides or
volcanic emissionsin emergency situations.Different
Lagrangian particle dispersion models have been developed
for those studies, like FLEXPART or HYSPLIT.It is very
important that the models are verified and their errors are
estimated so that the results are reliable.In this thesis
the advection and diffusion part of the new Lagrangian
particle dispersionmodel Massive Parallel Trajectory
Calculations (MPTRAC) was verified in idealized test cases
and inreal atmospheric conditions and optimized in terms of
accuracy and performance. Analytical test cases were applied
to the model to validate the advection.The implementation
and accuracy of integration schemes of different order,the
error caused by the linear interpolation of the wind fields
and the used coordinate system were discussed.All
implementations lead to correct and very accurate results. A
notable difference concerning the accuracy was only
determined between the first order Euler method and other
methods of higher order. Runge-Kutta methods of the order 2
to 4 and the Petterssen scheme led to very similar results,
indicating that an order of larger than 2 can not increase
the accuracy significantly.This is partly caused by the
linear interpolation of the meteorological data. Even if
tests have shown that the error caused by this linear
interpolation is small, higher order methods may need a
better interpolation to provide benefits.Tests for real
atmospheric conditions confirm these results, i.g., the
midpoint method was found to be the most performant
integration scheme. However, a small time step is required
to yieldsmall deviations from the reference solution. For
the tests with real atmospheric conditions the atmosphere
was separated into regions with similar conditions and the
resulting transport deviations are analyzed in detail.
Simulation errors in the stratosphere and in the tropics are
significantly smaller than in the troposphere or at high
latitudes.The diffusion scheme of MPTRAC consists of three
components, a random horizontal displacement, a random
vertical displacement, and a mesoscale diffusion that
depends on the variation of the wind field around an air
parcel.All three components were tested individually without
advection and produced results that correspond to analytical
solutions.However, it is difficult to predict a solution
under real atmospheric conditions, because the wind field
has a significant influence. A sensitivity test was done, to
get an overview of the impact of the diffusion components on
the transport of air parcels.The mesoscale diffusion has the
strongest influence, if default parameters are used,
followed by the vertical diffusion.The horizontal diffusion
has only a very small impact and does not seem to influence
the transport simulations significantly.In conclusion, the
model MPTRAC can be used to simulate the advection and
dispersion of air parcels correctly and
efficiently.Simulations with altitudes below the free
troposphere require relatively small time steps and the
diffusion model isprobably too simplistic, but the main task
of the model are simulations in the free troposphere and
stratosphere where the model shows good results.},
cin = {JSC},
cid = {I:(DE-Juel1)JSC-20090406},
pnm = {511 - Computational Science and Mathematical Methods
(POF3-511)},
pid = {G:(DE-HGF)POF3-511},
typ = {PUB:(DE-HGF)2},
url = {https://juser.fz-juelich.de/record/256610},
}