Buy Ebook Mathematical Problems in Meteorological Modelling 1st Edition András Bátkai Cheap Price

Full download test bank at ebook textbookfull.
com
Mathematical Problems in
Meteorological Modelling 1st
CLICK LINK TO DOWLOAD
https://textbookfull.com/product/mathematical
-problems-in-meteorological-modelling-1st-
edition-andras-batkai/
textbookfull
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
Mathematical Modelling in Real Life Problems Case

Studies from ECMI Modelling Weeks Ewald Lindner
https://textbookfull.com/product/mathematical-modelling-in-real-
life-problems-case-studies-from-ecmi-modelling-weeks-ewald-
lindner/
Mathematical Modelling of Waves in Multi Scale

Structured Media 1st Edition Alexander B. Movchan
https://textbookfull.com/product/mathematical-modelling-of-waves-
in-multi-scale-structured-media-1st-edition-alexander-b-movchan/
Mathematical Modelling in Solid Mechanics 1st Edition

Francesco Dell'Isola
https://textbookfull.com/product/mathematical-modelling-in-solid-
mechanics-1st-edition-francesco-dellisola/
Mathematical Modelling 1st Edition Seppo Pohjolainen

(Eds.)
https://textbookfull.com/product/mathematical-modelling-1st-
edition-seppo-pohjolainen-eds/
Mathematical Methods and Modelling in Applied Sciences
1st Edition Mehmet Zeki Sarikaya (Editor)
https://textbookfull.com/product/mathematical-methods-and-
modelling-in-applied-sciences-1st-edition-mehmet-zeki-sarikaya-
editor/
Modelling in Mathematical Programming: Methodology and

Techniques José Manuel García Sánchez
https://textbookfull.com/product/modelling-in-mathematical-
programming-methodology-and-techniques-jose-manuel-garcia-
sanchez/
Language and Culture in Mathematical Cognition Volume 4

Mathematical Cognition and Learning 1st Edition Daniel
B. Berch
https://textbookfull.com/product/language-and-culture-in-
mathematical-cognition-volume-4-mathematical-cognition-and-
learning-1st-edition-daniel-b-berch/
Exercises and Problems in Mathematical Methods of

Physics Giampaolo Cicogna
https://textbookfull.com/product/exercises-and-problems-in-
mathematical-methods-of-physics-giampaolo-cicogna/
Methods of Mathematical Modelling: Fractional

Differential Equations 1st Edition Harendra Singh
https://textbookfull.com/product/methods-of-mathematical-
modelling-fractional-differential-equations-1st-edition-harendra-
singh/
Mathematics in Industry 24
The European Consortium for Mathematics in Industry
András Bátkai
Petra Csomós
István Faragó
András Horányi
Gabriella Szépszó Editors
Mathematical
Problems in
Meteorological
Modelling
MATHEMATICS IN INDUSTRY 24
Editors
Hans Georg Bock
Frank de Hoog
Avner Friedman
Arvind Gupta
André Nachbin
Tohru Ozawa
William R. Pulleyblank
Torgeir Rusten
Fadil Santosa
Jin Keun Seo
Anna-Karin Tornberg
THE EUROPEAN CONSORTIUM

FOR MATHEMATICS IN INDUSTRY
SUBSERIES
Managing Editor
Michael Günther
Editors
Luis L. Bonilla
Otmar Scherzer
Wil Schilders
More information about this series at http://www.springer.com/series/4650
András Bátkai • Petra Csomós • István Faragó •
András Horányi • Gabriella Szépszó
Editors
Mathematical Problems
in Meteorological Modelling
123
Editors
András Bátkai Petra Csomós
Institute of Mathematics Institute of Mathematics
ERotvRos Lorand University ERotvRos Loránd University
Budapest, Hungary Budapest, Hungary
MTA-ELTE Numerical Analysis MTA-ELTE Numerical Analysis
and Large Networks Research Group and Large Networks Research Group
Budapest, Hungary Budapest, Hungary
István Faragó András Horányi

Institute of Mathematics Solymár, Hungary
ERotvRos Loránd University
Budapest, Hungary
MTA-ELTE Numerical Analysis
and Large Networks Research Group
Budapest, Hungary
Gabriella Szépszó
Hungarian Meteorological Service
Budapest, Hungary
ISSN 1612-3956 ISSN 2198-3283 (electronic)

Mathematics in Industry
ISBN 978-3-319-40155-3 ISBN 978-3-319-40157-7 (eBook)
DOI 10.1007/978-3-319-40157-7
Library of Congress Control Number: 2016954078
Mathematics Subject Classification (2010): 86A10, 65L06, 62H11
© Springer International Publishing Switzerland 2016

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology
now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, express or implied, with respect to the material contained herein or for any
errors or omissions that may have been made.
Printed on acid-free paper
This Springer imprint is published by Springer Nature

The registered company is Springer International Publishing AG Switzerland
Prediction is very difficult, especially if it’s
about the future.
Niels Bohr
Foreword
When I first began studying pure mathematics, I could never have foreseen that I
would spend the better part of my professional life in the area of meteorology. At
that time, designing a proof for some conjecture seemed much more appealing than
dealing with complex, high-dimensional weather forecasting models. Nevertheless,
the presence of rich dynamics for already low-dimensional dynamical systems
eventually piqued my curiosity for what the atmosphere, which could be considered
the ultimate chaotic dynamical system, might hold within it.
In the early 1990s, people started to give some thought on how to deal with
chaotic behaviour in the context of day-to-day weather forecasting. The opportunity
to contribute to an operational approach that would probe the predictability of the
atmosphere has been an exciting one. The Ensemble Prediction Systems, which at
the time were being developed at a few places worldwide, were all designed to
inform the forecasters about the occurrence of different possible weather regimes. It
was already a mathematical challenge to optimally select the model runs comprising
the ensemble. Given the computer capacity, it was only possible to conduct a
relatively small number of model runs, compared to the model phase dimension.
During the last two decades, predictability research has produced an important
shift from a deterministic to a probabilistic view in the field of weather and
climate forecasting. A closer connection between ensemble forecasting and data
assimilation naturally follows from this transition: the flow-dependent information
that ensembles provide regarding uncertainties for certain areas can be employed to
improve the use of observations in data assimilation, which in turn leads to better
performing ensembles. This hybrid data assimilation approach is a promising new
development that presents ample opportunities for mathematicians to excel, e.g. by
exploring the role of model error.
Without a doubt, the increase in computer performance has been a decisive factor
for the progress in meteorological modelling. As such, it is essential to be prepared
for the next generations of computer architectures, which will most certainly require
a high level of scalability for numerical models. Close interaction between computer
hardware and software experts, numerical scientists and modellers would therefore
seem to be a prerequisite to progress once exascale computers become the standard.
vii
viii Foreword
This book provides an attractive and varied introduction to topics where math-
ematicians and modellers from the meteorological community can cooperate with
and help one another. I sincerely hope it will stimulate mathematicians to engage
in solving problems which operational weather centres face now and in the near
future. I would also like to warmly acknowledge the editors and contributors for
their valued initiative and excellent work.
De Bilt, Netherlands Jan Barkmeijer

April 2016
Preface
Meteorological modelling, and in particular numerical weather prediction, belongs

to those frontiers of research where both mathematicians and meteorologists can
encounter striking new research problems. As a complex chaotic system, the atmo-
sphere can only be modelled using a large system of nonlinear partial differential
equations. Since theoretical investigations of these equations remain beyond our
reach, only numerical approximative solutions are feasible. However, great strides
have been made in the related mathematical disciplines in the last few years. The
corresponding keywords, like big data or the environment, are also central topics in
the European Union’s Horizon 2020 project.
These and many other aspects are what motivated us when, in May 2014,
we organized a workshop entitled “Mathematical Problems in Meteorological
Modelling” at Eötvös Loránd University in Budapest. The idea was to introduce
mathematicians to mathematical problems arising in meteorology and thus to spark
new collaborations and research. After the informative, stimulating talks by leading
experts, we decided to seek to reach a broader audience by publishing a volume
where selected problems could be highlighted. Many of the contributors to this
volume participated in or otherwise supported the conference. The main objective
of the present volume is to highlight the beauty of the development fields covered,
to demonstrate their mathematical complexity and more importantly to move
mathematicians to contribute to the further success of such practical applications
like weather forecasting and climate change projections.
Both the conference and this volume were made possible with the support of
the MTA-ELTE Numerical Analysis and Large Networks Research Group and the
European Consortium for Mathematics in Industry (ECMI). Founded more than
a quarter of a century ago, the ECMI is a consortium of academic institutions
and industrial companies that act cooperatively to promote and support the use of
mathematical models in any activity of social or economic importance. Although
traditionally meteorology has never been seen as an “industry”, it nevertheless
satisfies the previous definition, and one of the ECMI’s key priorities is to get
mathematicians involved in environmental applications.
ix
x Preface
We hope this volume can help reach this goal and reach many mathematicians.
We have divided the book into three parts, moving from mathematical and numerical
problems through air quality modelling to advanced applications in data assimilation
and probabilistic forecasting. We are confident that many mathematicians working
in numerical analysis, partial differential equations and stochastic analysis will find
a source of motivation and inspiration for their future research work.
Lastly, we of course wish to warmly thank the reviewers for their valued
contributions.
Budapest, Hungary András Bátkai

Budapest, Hungary Petra Csomós
Budapest, Hungary István Faragó
Solymár, Hungary András Horányi
Budapest, Hungary Gabriella Szépszó
April 2016
Contents
Part I Numerical Methods for Meteorological Problems

1 On a Conservative Finite-Difference Method for 1D
Shallow Water Flows Based on Regularized Equations .. . . . . . . . . . . . . . . 3
Alexander Zlotnik and Vladimir Gavrilin
1.1 Introduction .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 3
1.2 The 1d Shallow Water System of Equations,
Its Regularization and Discretization .. . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 4
1.3 Numerical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 12
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 17
2 Discretization in Numerical Weather Prediction:
A Modular Approach to Investigate Spectral
and Local SISL Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 19
Steven Caluwaerts, Daan Degrauwe, Fabrice Voitus,
and Piet Termonia
2.1 Introduction .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 19
2.2 Discretization of the Dynamical Core in NWP .. . . . . . . . . . . . . . . . . . . . 21
2.2.1 Time Discretization of the Dynamical Core .. . . . . . . . . . . . . . 22
2.2.2 Space Discretization of the Dynamical Core
and Grids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 25
2.3 Spectral SISL Discretization Schemes . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 29
2.4 Study of Local SISL Z-Grid Schemes. . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 31
2.4.1 The Z-Grid Approach . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 32
2.4.2 Geostrophic Adjustment of Z-Grid Schemes.. . . . . . . . . . . . . 33
2.4.3 SISL Z-Grid Schemes . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 38
2.4.4 Computational Aspects . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 40
2.5 Conclusion and Outlook . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 42
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 43
xi
xii Contents
3 Turbulence Modeling Using Fractional Derivatives . . . . . . . . . . . . . . . . . . . . 47

Béla J. Szekeres
3.1 Introduction .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 47
3.2 Preliminaries.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 48
3.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 48
3.3.1 The Fractional Newton’s Law of Viscosity .. . . . . . . . . . . . . . . 48
3.3.2 The Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 51
3.3.3 The Test Problem . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 51
3.4 Conclusion .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 53
Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 54
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 55
4 A Parallel Numerical Solution Approach for Nonlinear
Parabolic Systems Arising in Air Pollution Transport Problems . . . . . 57
János Karátson and Balázs Kovács
4.1 Introduction .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 58
4.2 The Numerical Solution Process and its Convergence . . . . . . . . . . . . . 59
4.2.1 Time Discretization .. . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 60
4.2.2 FEM Discretization in Space. . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 61
4.2.3 Outer Iteration: Damped Newton’s Method .. . . . . . . . . . . . . . 62
4.2.4 Inner Iteration: Preconditioned CG Method
Using Equivalent Operator Preconditioning.. . . . . . . . . . . . . . 63
4.3 Some Examples in Air Pollution Models . . . . . . .. . . . . . . . . . . . . . . . . . . . 67
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 69
Part II Air Quality Modelling

5 Eulerian and Lagrangian Approaches for Modelling
of Air Quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 73
Ádám Leelőssy, Tamás Mona, Róbert Mészáros, István Lagzi,
and Ágnes Havasi
5.1 Introduction .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 73
5.2 Eulerian Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 74
5.3 Lagrangian Models .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 78
5.3.1 Puff Models .. . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 80
5.3.2 Trajectory Models . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 80
5.4 Conclusion .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 82
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 83
6 Hydrodynamic Modeling of Industrial Pollutants
Spreading in Atmosphere .. . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 87
Vitaliy A. Prusov and Anatoliy Y. Doroshenko
6.1 Introduction .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 87
6.2 General Hydrodynamic Model of Air Circulation .. . . . . . . . . . . . . . . . . 89
6.3 A Model of Turbulence . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 92
6.4 A Model of Cloud and Precipitation . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 97
Contents xiii
6.5 Numerical Method for Solving the Non-stationary

Problem with a Prehistory Based on Interpolation
with Multiple Nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 103
6.6 Interpolating Functions, Specified in Macro-Scale Grid
Nodes, into the Meso-Scale Grid . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 107
6.7 Approximation of Constituent Members
of the Convection-Diffusion . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 109
6.8 Solution of the Problem of Impurities Dispersion
in the “Near Field” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 112
6.9 Conclusion .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 114
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 115
7 Coordinate Transformation Approach for Numerical
Solution of Environmental Problems .. . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 117
Tatiana Chernogorova, Ivan Dimov and Lubin Vulkov
7.1 Introduction .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 117
7.2 A Stationary Model of Air Pollution . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 119
7.3 A Non-stationary Two-Dimensional Problem . .. . . . . . . . . . . . . . . . . . . . 122
7.4 Positive Splitting Numerical method .. . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 124
7.5 Conclusions .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 126
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 127
8 Impact of Climatic Changes on Pollution Levels . . . .. . . . . . . . . . . . . . . . . . . . 129
Zahari Zlatev, István Faragó, and Ágnes Havasi
8.1 Introduction .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 130
8.2 Development of Three Climatic Scenarios . . . . .. . . . . . . . . . . . . . . . . . . . 131
8.2.1 Climate Scenario 1 (Taking into Account
Only the Future Changes of the Temperatures) . . . . . . . . . . 131
Some Additional Factors) . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 132
More Additional Factors) . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 133
8.2.4 Some Remarks Related to the Climatic Scenarios . . . . . . . . 133
8.3 Presentation of the Mathematical Model . . . . . .. . . . . . . . . . . . . . . . . . . . 134
8.4 Additional Computational Difficulties Arising in the
Climatic Runs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 137
8.5 Inter-Annual Variations of the Meteorological
Conditions and Some Other Issues . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 146
8.5.1 Why Are the Inter-Annual Variations of the
Meteorological Conditions Important? . . . . . . . . . . . . . . . . . . . . 146
8.5.2 Need for Scenarios with Human-Made and
Natural Emissions . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 147
8.6 Influence of the Climatic Changes on Pollution Levels . . . . . . . . . . . . 150
8.7 Conclusions .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 158
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 159
xiv Contents
Part III Meteorological Data Assimilation and Probabilistic

Forecasting
9 An Invitation to Meteorological Data Assimilation .. . . . . . . . . . . . . . . . . . . . 165
Ágnes Bodó and Petra Csomós
9.1 Introduction .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 165
9.2 Mathematical Background .. . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 166
9.3 Optimal Interpolation and Variational Method in One
Dimension . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 169
9.4 Optimal Interpolation and Variational Method in More
Dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 172
9.5 Kálmán Filter Techniques .. . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 177
9.6 Numerical Experiments . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 183
9.6.1 Linear Iteration . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 183
9.6.2 Lorenz System . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 185
9.7 Outlook: Nonlinear Data Assimilation . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 190
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 191
10 Analysis of the Data Assimilation Methods
from the Mathematical Point of View . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 193
Tamás Szentimrey
10.1 Introduction .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 193
10.2 Data Assimilation Model Applied in the
Meteorological Practice . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 195
10.2.1 The Cost function of Data Assimilation .. . . . . . . . . . . . . . . . . . 195
10.2.2 Problems with the Cost Function . . . . . .. . . . . . . . . . . . . . . . . . . . 195
10.2.3 Problems with the Reanalysis Data Based
on Data Assimilation . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 196
10.3 Derivation of the Bayesian Cost Function for Data
Assimilation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 196
10.3.1 Application of the Bayes Theorem . . . .. . . . . . . . . . . . . . . . . . . . 196
10.3.2 The Normal Conditional Density Functions
and their Properties . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 197
10.3.3 The General Form of the Cost Function .. . . . . . . . . . . . . . . . . . 198
10.4 The Properties of the General Cost Function . . .. . . . . . . . . . . . . . . . . . . . 199
10.4.1 Examination of E .X j Xb D xb/
and B D Cov ŒX E .X j Xb/ . . . . . . . .. . . . . . . . . . . . . . . . . . . . 199
10.4.2 Examination of the Structure of Cost Function . . . . . . . . . . . 201
10.5 The Problem of Conditional Expectation
of Atmospheric State, Given Forecasts .. . . . . . . . .. . . . . . . . . . . . . . . . . . . . 202
10.6 Summary and Conclusion.. . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 204
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 204
Contents xv
11 Ensemble Methods in Meteorological Modelling . . .. . . . . . . . . . . . . . . . . . . . 207

Mihály Szűcs, András Horányi, and Gabriella Szépszó
11.1 Introduction .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 208
11.2 Uncertainties in Numerical Weather Predictions
and Climate Projections .. . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 208
11.3 Ensemble Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 209
11.3.1 Initial Condition Perturbations .. . . . . . . .. . . . . . . . . . . . . . . . . . . . 211
11.3.2 Representation of Model Uncertainties .. . . . . . . . . . . . . . . . . . . 218
11.3.3 Representation of Uncertainties Related
to Anthropogenic Activity . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 222
11.3.4 Other Methods . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 224
11.4 Applications of Ensemble Forecasts . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 224
11.4.1 Some Examples of Ensemble Systems . . . . . . . . . . . . . . . . . . . . 225
11.4.2 Visualization Methods .. . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 229
11.5 Summary.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 233
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 235
12 Quantifying Sources of Uncertainty in Temperature
and Precipitation Projections over Different Parts of Europe .. . . . . . . . 239
Péter Szabó and Gabriella Szépszó
12.1 Introduction .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 240
12.2 Methods .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 242
12.3 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 244
12.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 246
12.4.1 Future Temperature and Precipitation Trends .. . . . . . . . . . . . 246
12.4.2 Internal Variability .. . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 249
12.4.3 Fractions of Uncertainty.. . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 251
12.4.4 Potentially Reducible Uncertainty . . . . .. . . . . . . . . . . . . . . . . . . . 253
12.4.5 Signal-to-Noise Ratio and Time of Emergence . . . . . . . . . . . 255
12.5 Summary.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 258
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 260
Index . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 263
Part I
Numerical Methods for Meteorological
Problems
The dynamics of the atmosphere is governed by physical, chemical, and even

biological processes which are commonly described by systems of time and space
dependent nonlinear partial differential equations. Since this kind of mathematical
description is rather complicated, the form of the equations’ exact solution is usually
unknown. In order to explore its properties or to compute its approximation, further
mathematical methods are needed. The resulting atmospheric models are then to
forecast the weather situation, the concentration of an air pollutant, or even the
changes in climate. Since the computation of the forecast should not miss the
events occurred in reality, it requires even faster and of course more accurate
numerical treatment. These requirements stimulate the researches to elaborate on
new numerical methods.
This part of the issue is devoted to the introduction and analysis of the basic ideas
and methods behind atmospheric modelling. The chapters give an insight to the most
commonly used numerical techniques. In order to distinguish the important features
of the numerical methods themselves from the ones caused by the processes being
described, the authors apply the methods to the same mathematical problems which
arise in meteorology, however, in their simplified forms. Only after having showed
the nice properties of a certain method, it can be applied to the more complex
atmospheric problems presented in Parts II and III of the issue.
In Chap. 1, the authors analyse finite difference schemes being the most basic
time discretisation method used when solving partial differential equations. They
apply them to the one-dimensional shallow water equations which served as the
first reliable model for large-scale atmospheric processes.
Chapter 2 introduces another discretisation method used in meteorological
modelling. After giving an introduction on the mostly applied space discretisation
methods, the authors focus on the combination of the semi-implicit finite difference
method in time with the semi-Lagrangian approach for the nonlinear advection part,
and apply it to two-dimensional shallow water equations.
Chapter 3 gives an introduction to the modelling of two-dimensional turbulence
by using fractional derivatives. The main idea is to treat the shear stresses as random
variables. The author present an algorithm, and apply it to a test problem.
2 I Numerical Methods for Meteorological Problems
In Chap. 4, the authors derive and show the convergence of a parallel numerical
method for general nonlinear parabolic problems which usually arise in air pollution
transport problems. Due to the chemical part, they face with nonlinear elliptic
problems.
As seen from this short summary, Chaps. 1 and 2 treat numerical methods
being the right choice when modelling the large-scale dynamics of the atmosphere,
Chap. 3 deals with the modelling of turbulence occurring on much smaller scales,
and Chap. 4 gives a generic treatment of problems arising in, for instance, air quality
modelling. All of these fields will be covered in Parts II and III of this issue.
Chapter 1
On a Conservative Finite-Difference Method
for 1D Shallow Water Flows Based
on Regularized Equations
Alexander Zlotnik and Vladimir Gavrilin
Abstract We deal with the 1d shallow water system of equations and exploit its
special parabolic regularization satisfying the energy balance law. We construct a
three-point symmetric in space discretization such that the discrete energy balance
law holds and check that it is well-balanced. The results of numerical experiments
for the associated explicit finite-difference scheme are also given for several known
tests to confirm its reliability and some advantages. The practical error behavior is
also analyzed.
Keywords 1D shallow water equations • Discrete energy balance law • Parabolic

regularization • Practical error analysis • Three-point symmetric in space
discretization
1.1 Introduction
The 1d and 2d shallow water equations are widely exploited in hydrodynamics. The
vast literature exists on their numerical solving, in particular, see [14] and [1, 3, 12,
15, 17–19], etc.
Several years ago a special regularization of the equations together with explicit
two-level in time and symmetric in space finite-difference schemes based upon it
were suggested in [5, 8]. The approach is further applied for various problems
in [4, 9–11]. Previously the similar approach was developed for the more com-
plicated Euler and Navier–Stokes equations and its practical efficiency in various
A. Zlotnik ()
Department of Higher Mathematics at Faculty of Economic Sciences, National Research
University Higher School of Economics, Myasnitskaya 20, 101000 Moscow, Russia
e-mail: [email protected]
V. Gavrilin
Department of Mathematical Modelling, National Research University Moscow Power
Engineering Institute, Krasnokazarmennaya 14, 111250 Moscow, Russia
e-mail: [email protected]
© Springer International Publishing Switzerland 2016 3

A. Bátkai et al. (eds.), Mathematical Problems in Meteorological Modelling,
Mathematics in Industry 24, DOI 10.1007/978-3-319-40157-7_1
4 A. Zlotnik and V. Gavrilin
compressible gas flow computations was demonstrated [6, 7, 16]; it is called a quasi-
gas dynamic (QGD) approach. Moreover, the regularized shallow water system of
equations is nothing more than a special case of the main version of the barotropic
QGD system, see [20, 22].
In this chapter, following [21–23], we change the space discretization in the
above mentioned schemes in order to guarantee the discrete energy balance law
not available originally. We present the proof of the law and check the important
property that the new discretization is well-balanced. The results of numerical
experiments for the associated explicit in time finite-difference scheme are also
given. We treat known tests such as a dam break (a disintegration of discontinuity)
and subcritical, transcritical and supercritical flows over a hump to confirm the
reliability and accuracy compared with the original scheme. In addition, the
practical error behavior is analyzed as mesh refines. Here we confine ourselves by
the 1d case; the 2d case should be considered in a forthcoming paper.
1.2 The 1d Shallow Water System of Equations,

Its Regularization and Discretization
1.2.1. The 1d shallow water system of equations consists of the mass and momentum
balance equations
@t h C @x .hu/ D 0; (1.1)
@t .hu/ C @x .hu2 / C @x p D @x ˘NS C hF; (1.2)
where 0 < x < X, t > 0 and @t and @x denote the time and space derivatives. The
main unknown functions are the water depth h.x; t/ > 0 (measured from the bottom
mark b.x/) and the velocity u.x; t/. The pressure, the Navier–Stokes viscous stress
and the density of external force are given by
p D 0:5gh2 ; ˘NS D @x u; F.x; t/ D g@x b.x/ C f .x; t/; (1.3)
where g D const > 0, D .h/ > 0 is the viscosity coefficient (when D 0,

viscosity is ignored) and f is a perturbation of the stationary force g@x b. Remind
also the important quantities H D h C b (the water level) and hu (the discharge).
The equations are supplemented with the initial conditions
h.x; 0/ D h0 .x/; u.x; 0/ D u0 .x/; 0 6 x 6 X: (1.4)
The regularized 1d shallow water system of equations consists of the modified

mass and momentum balance equations
@t h C @x .h.u w// D 0; (1.5)
@t .hu/ C @x .h.u w/u C p/ D @x ˘ C h F: (1.6)

1 On a Conservative Finite-Difference Method 5
Here the density of the mass flux j, the viscous stress ˘ and the regularized density
h are given by

j D h.u w/; w D wO C u@x .hu/; wO D .hu@x u C @x p hF/; (1.7)
h h
˘ D @x u C huw
O C gh@x .hu/; h D h @x .hu/; (1.8)
where D .h; u/ > 0 is the relaxation multiplier. If one puts D 0, the system
becomes the original system (1.1)–(1.3).
Note that the following simplified formula holds
˚
wO D u@x u C g@x .h C b/ f : (1.9)
The regularized system (in equivalent form) was derived from the original one
in [5, 8] in 1d and 2d cases. On the other hand, the latter system is nothing more
than the special case of the main version of the barotropic QGD system of equations
from [20] (its earlier version can be found in [24]) for the gas density D h and the
pressure law p./ D 0:5g2 with the adiabatic exponent D 2. In 1d this is due to
formulas (1.9) and .u@x p C p@x u/ D gh@x .hu/.
For the barotropic QGD system of equations, the Petrovskii parabolicity, the
energy balance law and the linearized stability of the equilibrium solutions were
proved in [20, 22]; its very short derivation can be also found there. Remind that,
first, the original system (1.1) and (1.2) is hyperbolic for D 0 or of the composite
hyperbolic-parabolic type for > 0 and, second, the Petrovskii parabolicity implies
a unique solvability (local in time) for the Cauchy problem in Hölder spaces, see
[24] for details. These results demonstrate the dissipative nature of the barotropic
system. They straightforwardly imply the corresponding results for the regularized
shallow water system of equations, in particular, the following important one.
Theorem 1.1 For the regularized shallow water equations (1.5)–(1.8), the follow-
ing pointwise energy balance law holds
˚ ˚
0:5@t g.h C b/2 C hu2 C @x h.u w/ g.h C b/ C 0:5u2 ˘ u
˚ 2 ˚ 2
C.@x u/2 C g @x .hu/ C h u@x u C g@x .h C b/
˚
D h fu C h u@x u C g@x .h C b/ f : (1.10)
In the law on the left, the second term is a divergence one whereas the third
(Navier–Stokes) term and fourth and fifth relaxation terms are non-negative; the
property remains valid for > 0.
For f D 0 and suitable boundary conditions at x D 0; X, for example, for the

periodic ones, clearly law (1.10) implies non-increasing of the total energy in time
Z X˚
0:5@t g.h C b/2 C hu2 dx 6 0: (1.11)
0
Notice that, for the equilibrium solutions h D hS .x/ > 0 and u D 0, both
systems (1.1)–(1.3) and (1.5)–(1.8) for f D 0 are reduced to the equation
@x p.hS / C ghS @x b D 0 on .0; X/; (1.12)
i.e., simply to the formula hS .x/ C b.x/ C on Œ0; X.

1.2.2. We define a non-uniform mesh !N h in space with nodes 0 D x0 < x1 <
< xN D X and steps i D xi xi1 . Let max WD max16i6N i . We also use an
auxiliary (conjugate) mesh !h with nodes xiC1=2 D .xi C xiC1 /=2, 0 6 i 6 N 1,
and steps O i D xiC1=2 xi1=2 D .i C iC1 /=2.
For functions v given on !N h and y given on !h , we introduce averages, shifts of
argument and difference quotients
viC1 vi
ŒviC1=2 D 0:5.vi C viC1 /; v;iC1=2 D vi ; vC;iC1=2 D viC1 ; ıviC1=2 D ;
iC1
i iC1 yiC1=2 yi1=2
Œ yi D yi1=2 C yiC1=2 ; ı yi D :
O
2i O
2 i O i
The operators Œ, ./˙ and ı map functions given on !N h to ones defined on !h
whereas Œ and ı map functions given on !h to those defined on !h D fxi giD1
N1
.
We need two difference product rules
ı.uv/ D ıu Œv C Œuıv; (1.13)
ı .yŒv/ D ı y v C Œ yıv (1.14)
(u is given on !N h ). To reduce the amount of brackets, hereafter we suppose that,

for example, ıu Œv D .ıu/Œv (i.e., the multiplication sign terminates action of
preceding operators). Formula (1.14) was effectively applied, in particular, in [2].
We also exploit the formulas
viC1 vi1
Œıvi D ı Œvi D ; (1.15)
2O i

Œ y v D yŒv 0:25ı .2C yıv/: (1.16)
Turning to [23], we first construct a three-point symmetric in space discretization

to (1.5)–(1.8) and write down the semi-discrete mass and momentum balance
equations as follows

@t h C ı Œh.Œu w/ D 0; (1.17)

@t .hu/ C ı Œh.Œu w/Œu C Œ p D ı ˘ C Œh F (1.18)
on !h for t > 0. Here the following discretizations for the density of the mass flux
j, the viscous stress ˘ and the regularized density h are exploited

j D Œh.Œu w/; w D wO C Œuı.hu/; wO D ŒhŒuıu C ıp ŒhF ;
Œh Œh
(1.19)
˘ D ıu C ŒhŒuwO C gŒhı.hu/; h D Œh ı.hu/; F D gıb C f : (1.20)
The main unknown functions h and u together with the functions p and b are defined
on the main mesh !N h whereas the functions j, w, w,
O ˘ , h , , and f are defined
on the auxiliary mesh !h . Note that ıp D gŒhıh and therefore similarly to (1.9) the
following formula holds
O D fŒuıu C gı.h C b/ f g:
w (1.21)
We assume that h > 0 and derive the counterpart of the energy balance
law (1.10).
Theorem 1.2 For the semi-discrete method (1.17)–(1.20), the following pointwise
energy balance law holds
˚ ˚
0:5@t g.h C b/2 C hu2 C ı j .g.h C b/ C 0:5u uC / ˘ Œu C B
˚ 2
C .ıu/2 C g ı.hu/ C ŒhfŒuıu C gı.h C b/g2

D Œh f u C ŒhfŒuıu C gı.h C b/gf ; (1.22)
where B WD 0:252C .ıp C gh ıb/ıu.

In the law on the left, all three terms under the averaging sign Œ are non-
negative; the property remains valid for > 0.
Proof We follow the derivation of the differential law (1.10) from [22], see also
[23], and first multiply the mass balance equation (1.17) by g.h C b/. Since

ı j g.h C b/ D ı . jgŒh C b/ jgı.h C b/
according to formula (1.14), we get

@t . p.h/ C ghb/ C ı . jgŒh C b/ Œhgı.h C b/.Œu w/ D 0: (1.23)
Next we multiply the momentum balance equation (1.18) by u. We apply the

formula
@t .hu/ u D 0:5@t .hu2 / C 0:5@t h u2 ;
the mass balance equation (1.17) and twice formula (1.14) to obtain:

@t h u2 D ı j u2 D ı . jŒu2 / C jı.u2 / ;

ı . jŒu/ u D ı . jŒu2 / jŒuıu :
Then also taking into account elementary formulas
Œu2 D 0:5Œu2 C 0:5u uC ; 0:5ı.u2 / D Œuıu (1.24)
and formula (1.15), we derive
0:5@t .hu2 / C 0:5ı . ju uC / C Œıp C gh ıb u ı ˘ u D Œh f u:
We add equality (1.23) and the last one. Formulas (1.14) and (1.16), respectively,
imply
˚
Œıp C gh ıb u D ıp C gŒhıb ı.hu/ gıb Œu C ı B ;
ı ˘ u D ı .˘ Œu/ C Œ˘ ıu ;
with the above introduced B . Since ıp C gŒhıb D Œhgı.h C b/, we obtain

˚
0:5@t g.h C b/2 C hu2 C ı .A C B /

C Œhgı.h C b/w Œuı.hu/ gıb C ˘ ıu D Œh f u;
with A WD j fg.h C b/ C 0:5u uC g ˘ Œu.

Extracting the terms f in the multipliers w and ˘ , we rewrite the last equality
in the form
˚
0:5@t g.h C b/2 C hu2 C ı .A C B / C .ıu/2 C .h; u/

D Œh f u C fŒhgı.h C b/ C ŒhŒuıugf ; (1.25)
where
.h; u/ WD Œhgı.h C b/.w C f / Œuı.hu/ gıb
CfŒhŒu.wO C f / C gŒhı.hu/gıu
D Œhgı.h C b/.w C f / C ŒhŒuıu .wO C f /
gıb Œuı.hu/ C gŒhı.hu/ ıu:
Applying the formulas

˚
w C f D wO C f C Œuı.hu/; w
O C f D Œuıu C gı.h C b/ ;
Œh
see (1.21), we transform .h; u/ as follows
.h; u/
˚
D Œhgı.h C b/ C ŒhŒuıu .wO C f / C gıh Œuı.hu/ C gŒhı.hu/ ıu
˚ 2 ˚
D Œh Œuıu C gı.h C b/ C ı.hu/ gıh Œu C gŒhıu :
Finally, using the formula ıh Œu C Œhıu D ı.hu/, see (1.13), we rewrite .h; u/ as
the sum of two squared terms and thus pass from (1.25) to the desired result (1.22).
t
u
For f D 0 and suitable boundary conditions at x D 0; X, for example, for the
periodic ones, clearly law (1.22) implies non-increasing of the total energy in time
X
N1
˚
0:5@t g.h C b/2 C hu2 i O i 6 0;
iD0
where O 0 D .h1 C hN /=2, that is the discrete counterpart of the energy inequal-
ity (1.11).
In the discrete energy balance law (1.22), the term u uC is like the geometric
mean for u2 . The summand ı B is the divergence mesh imbalance [additional with
respect to (1.10)] with B D O.2max / for the functions of the continuous argument
provided that h, @x h, @x u and @x b are bounded.
For the equilibrium solution h D hS .x/ > 0 and u D 0, method (1.17)–(1.20) for
f D 0 is reduced to the equations [with the help of formula (1.15)]
˚
ı ıp.hS / C gŒhS ıb D 0; ıp.hS / C gŒhS ıb D 0 on !h :
Notice that ıp.hS / C gŒhS ıb D gŒhS ı.hS C b/. By virtue of the first equation
gŒhS ı.hS C b/ D C0 const on !h , and next C0 D 0 due to the second equation.
Consequently
gŒhS ı.hS C b/ D 0 on !h ;

i.e. hS C b const on !N h . These results correspond to (1.12) and its consequence

for the regularized system and mean that the method is well-balanced which is the
known important property.
1.2.3. Let us compare our discretization and one from [5, 8, 11], for which
unfortunately the energy balance law is not known. The following two terms in
the momentum balance equation (1.18)

ı Œ p.h/; Œh F D fŒh ı.hu/g.gıb C f /
differ from those

˚
ı p.Œh/; Œh Q ı.ŒhŒu/ .gı Œb C fQ /;
in [5, 8, 11] (generalized for the non-uniform !N h and written in our notation), where
Q and fQ are defined on !N h . Moreover, inserting formula (1.21) into the second
formula (1.19) leads to
˚
wD 2ŒhŒuıu C Œu2 ıh C gı.h C b/ f
Œh
whereas in [5, 8, 11] a formula equivalent to
Œ
Q ˚
wD Q gı.h C b/ fQ
ı.hu2/ C Œ
Œh
was applied. Since ı.hu2 / D 2ŒhŒuıu C Œu2 ıh but Œu2 6 Œu2 [see (1.13)
and (1.24)], these two formulas for w also differ.
On the other hand, inserting formula (1.21) into the first formula (1.20) gives
˚
˘ D ıu C ŒhŒu ı g.h C b/ C 0:5u2 f C gŒhı.hu/
similarly to [5, 11] (up to expressions for , and f ).

1.2.4. Now we define a non-uniform mesh in time with nodes 0 D t0 < : : : <
tM D tfin (which are not prescribed a priori) and steps tm D tmC1 tm , where tfin is
the given final time.
We approximate the time derivatives in the mass and momentum equations (1.15)
and (1.16) by the forward differences and get the explicit finite difference scheme

hO D h t ı Œh.Œu w/ ; (1.26)
˚
b D hu t ı Œh.Œu w/Œu C Œ p ˘ Œh F
hu (1.27)
on !h for 0 6 m 6 M 1, where hO and hu b are the values for the upper time level.
After computing them, we set uO WD hu=h assuming that hO > 0. Here we also exploit
b O
the same expressions (1.19) and (1.20).
For solving the inviscid shallow water system of equations, the viscous and
relaxation terms are considered as artificial regularizers with and in the form
4 p
D Œ p or D 0; D ˛ ; c D gŒh
3 c
on !h like in [7, 16], where c is the counterpart for the velocity of sound and 0 <
˛ < 1. To satisfy a stability condition of the Courant-Friedrichs-Lévy type, the time
step at the current time level is chosen as

t D ˇ min
16i6N jŒui1=2 j C ci1=2
with 0 < ˇ < 1, and thus t varies from one time level to another. The scheme
parameters ˛ and ˇ are adjusted experimentally in each problem.
Notice that, at least for the uniform space mesh, the approximation order of the
semi-discrete method (1.17)–(1.20) [with respect to the regularized system (1.5)–
(1.8)] is O.2 / with D X=N for smooth solutions but it is reduced to O./ for
the finite-difference scheme owing to the above choices of and t.
1.2.5. In our computations below, we use a uniform mesh in space and take f D 0.
We investigate the convergence of functions v D h; u and hu for our finite difference
scheme at the final time tfin as the space mesh refines; this is a standard matter,
for example see [13, 15, 17], and especially important in the case of non-smooth
solutions where the theoretical information is poor. In those rare cases where a
sought function vex is known, we compute the average absolute value of the error
on the mesh (the scaled mesh L1 -norm)
1 Xˇˇ .N/ ˇ
N
.N/
Eex Œv WD vex i vi ˇ; (1.28)
N C 1 iD0
.N/
where vex i and vi are the exact and approximate quantities at xi D iX=N.
Otherwise, we apply two formulas
1 Xˇˇ .N/ .N/ ˇ 1 Xˇˇ .2N/ .N/ ˇˇ

N N
viN=N vi ˇ; ER Œv WD
.N/
E.N/ Œv WD v vi ; (1.29)
N C 1 iD0 N C 1 iD0 2i
where v .N/ is the pseudo-exact solution computed for the finer mesh with the step
D X=N. Here N is a multiple of 2N, and due to our numerical experience it is
desirable that 4N 6 N. The second formula is associated to the Runge rule for the
practical error estimation and is cheaper in implementation (v .N/ is not required).
We compare both of them with (1.28) if possible.
ˇ ˇ
Below we plot the dependence of ˇln E.N/ Œvˇ (the subscripts are omitted) on ln N.
Moreover, we approximate them by linear functions using the least square method
and thus get the corresponding approximate error orders pex Œv, pŒv and pR Œv.
1.3 Numerical Results
1.3.1. We first compute a flow in a channel of the length X D 1500 m. In its central
part, a rectangular ledge is symmetrically located (having the height 8 m and the
length 375 m), and the other bottom is flat. At the center of the channel, a dam is
situated. Initially the water level H0 .x/ equals 20 m to the left and 15 m to the right
of the dam, and the flow is at rest, i.e. the initial velocity is zero: u0 .x/ D 0. The
dam breaks at the moment t D 0. The open boundary conditions are imposed for h
and u at the both boundaries x D 0; X, i.e., for example, h0 D h1 and u0 D u1 for
the left boundary.
The computational results are shown in Fig. 1.1 for t D 15 s and t D 60 s, for
N D 400; the scheme parameters are ˛ D 0:35 and ˇ D 0:5. They are in a good
agreement with [15] for N D 200 and in [11] for N D 400 (for other schemes).
H u
25 3
2.5
20
10 1.25
0 0
100 200 300 400 100 200 300 400
H u
25
3.5
20
2.5
10 1.5
0.5
0
100 200 300 400 100 200 300 400
Fig. 1.1 Dam break in the channel with the rectangular ledge: H and u at t D 15 s (above) and
t D 60 s (below)
4.5 E(h)
E(u)
| ln E |
3.5
2.5
5.25 5.5 6 6.5 7 7.5

ln N
ˇ ˇ
Fig. 1.2 Dam break in the channel with the rectangular ledge: ˇln E.N/ ˇ for h and u
In Fig. 1.2, the errors in h and u for t D 15 s are presented for N D 200, 400, 800
and 1600 as well as N D 6400. We note that the error polylines are not completely
linear. The error orders pŒh 0:946 and pŒu 0:908 are rather close to each other
but both less than 1.
1.3.2. Next we consider a channel of the length X D 25 m with a flat bottom
except for the small hump of the parabolic shape in its middle part:
(
0:2 0:05.x 10/2 ; 8 m x 12 m
b.x/ D
0; otherwise:
Initially the water level is constant: H0 .x/ CH , and the flow is at rest. The left
boundary conditions are hujxD0 D Chu for the discharge together with the open
boundary condition for h, and the right boundary conditions are HjxDX D CH (in
general, up to a certain time moment) together with the open boundary condition
for u.
This problem may seem simple but only at first glance. There are three types of
flows in it: subcritical, transcritical and supercritical depending on values of the flow
parameters. In what follows, examples of all three types of flows are considered.
Results of computations are presented at tfin D 200 s (at this moment the flows
become stationary for the chosen values of the parameters). Note that, for these
flows, the exact discharge at the final time is known: hu Chu . As a rule, namely
the computation of hu causes difficulties. The results are accurate enough for N
listed below; they are comparable with those obtained in [3] (for N D 100), [12]
(for N D 300) and [1, 18, 19] (for N D 200) by other schemes but much better for
hu than in [11] (for the same N).
(a) Subcritical flow. This is the simplest type of flows. Here CH D 2 m and Chu D
4:42 m2 /s are taken. At tfin , the water level is almost flat with only a small cavity
above the hump.
We select ˛ D 0:9, ˇ D 0:2 and D 0. Figure 1.3 shows the water level H
and the discharge hu for N D 400. In the vicinity of the hump edges, “hubbles”
H hu
2.5 4.422
4.42
1
0 4.418
100 200 300 400 100 200 300 400
Fig. 1.3 Subcritical flow over the hump: H and hu at tfin D 200 s
14
E(h)
E(u)
11
E(hu)
E (h)
| ln E |
R
E R (u)
E R (hu) 8
E ex (hu)
4.5 5 5.5 6 6.75

ln N
ˇ ˇ
Fig. 1.4 Subcritical flow over the hump: ˇln E.N/ ˇ for h, u and hu
of hu values are observed (damping as N increases) but the absolute error in the
discharge Eabs WD max j.hu/i Chu j 7:375e-5 is small.
0iN
In Fig. 1.4, errors in h, u and hu are shown for N D 100, 200, 400 and 800 as
well as N D 3200. The same values are taken for two other flow types below.
The polylines of errors in h and u are more close to linear ones than for the
previous problem (though this is not the case for jln Eex Œhuj). Now all the orders
pŒh 1:193, pR Œh 1:139, pŒu 1:183 and pR Œu 1:121 are all close to each
other and slightly greater than 1. They are essentially less than pex Œhu 1:709.
Comparing the results by formulas (1.28) and (1.29) for hu, we see that the error
orders pR Œhu 2:023 and pŒhu 2:017 are close to each other but both
overestimate the value pex Œhu.
(b) Transcritical flow. Here CH D 0.66 m and Chu D 1:53 m2 /s are chosen. For this
and the next types of flows, the right boundary condition H.X; t/ D CH is posed
only for t 40 s whereas it is replaced by the open boundary condition for h
for t > 40 s. In this case, the behavior of the stationary water level H is more
complicated exhibiting much more sharp change (its values on the left and right
Another random document with
no related content on Scribd:
noteworthy—there are no altar rails, but their place is taken by three
massive oaken benches, covered at all times with the “houseling
linen,” fair white cloths—the use of which goes back to very early
times. The benches which now stand across the presbytery floor,
close to the topmost of the flight of steps leading up to it, are the
remains of ten such benches, which were made in Puritan times for
communicants to sit on as they received the sacred elements. When
the custom of kneeling was revived, these benches were placed on
the steps; and on “Sacrament Sundays,” the clerk, after morning
prayer, went to the lectern and bade those who were prepared to
receive the Holy Communion to draw near, whereupon intending
communicants left the nave and knelt at the benches, or in the choir
stalls, until the officiating clergy brought them the sacramental bread
and wine. In 1852, when sundry changes were being made in the
arrangements of the church, all these benches except three were
removed—the three which were in use as altar-rails.
The beautiful triplet of windows, over the altar, end the long vista
seen from the west. Beneath the presbytery floor is a vaulted crypt.
This is not, as many crypts are, dark and gloomy, but well lit by
triangular windows, which from the outside are seen to be just above
the level of the churchyard. This crypt does not retain its original
altar, but its place is marked by a piscina on the southern side; two
arches open out into the choir aisles, through which those kneeling
in the aisles might look down on the priest officiating at the altar of
the crypt.
There are several monuments worthy of notice, but by far the most
interesting is the Beaufort altar-tomb on the south side of the
presbytery. This was erected, to the memory of her father and
mother, by the Lady Margaret. foundress of Christ’s and St. John’s
Colleges, at Cambridge, well known for her many benefactions, and
from the fact that she was the mother of Henry VII. As the history of
this family—the Beauforts—is interesting, and its details little
touched on by the writers of English history, it may not be out of
place, especially as they were connected with Wimborne, to give an
outline of it here. Everyone knows that John of Gaunt was the third
son of Edward III. and the father of Henry IV., but many know only in
a general way that the House of Tudor traced through him their claim
to the English crown. John of Gaunt married Blanche, great-
granddaughter of Edmund Crouchback, second son of Henry III.,
who was created Earl of Lancaster in 1257. Her father, on account of
his valour in the French wars of Edward III., had been made a Duke
—a new title as far as England was concerned, for the only English
Duke that had been previously created was the Duke of Cornwall,
better known as the Black Prince. Through Blanche, his wife, John of
Gaunt succeeded to the estates of the Duke of Lancaster, among
them to the Castle of Beaufort, in Anjou. He was himself in 1362
created Duke of Lancaster. Among Blanche’s maids of honour was
one Kate, daughter of Sir Payne Roet, and widow of Sir Owen
Swynford. When Blanche died, John married Constance of Castile,
but took unto himself Kate Swynford as his mistress; by her he
became the father of four children, all born at Beaufort Castle. As
they were illegitimate, they took the name of their birthplace as a
surname. The eldest of these was John de Beaufort, and the
second, Henry, the celebrated Cardinal Beaufort. When Constance
died, these four children were legitimatised by a Bull of Urban VI.,
then by Richard II., then by Act of Parliament. The Duke then
married Kate. On January 13th, 1396, John de Beaufort was created
Earl of Somerset. He died in 1410, leaving four sons and two
daughters; the eldest, Henry, did not long survive his father, and his
title and estates passed to his next brother, John. He greatly
distinguished himself in the French wars, and was made Duke of
Somerset, Earl of Kendale, Lieutenant of the Duchy of Aquitaine and
Captain-General of the whole realm of France and Normandy. In
1436 the Duke of Bedford, the Regent, died, and as the King, Henry
VI., was still a minor, another regent had to be appointed. The Duke
of Lancaster thought he should have obtained this important post,
but it was conferred on the Duke of York, and Lancaster therefore
retired from active service, and in 1440 married Margaret, widow of
Oliver St. John, and daughter of John, Lord Beauchamp, of Bletsoe
Manor, Bedfordshire. This John, Duke of Somerset, and Margaret,
his wife, are they whose figures lie side by side in alabaster on their
altar-tomb at Wimborne. Their right hands are clasped together;
angels guard their heads; his feet rest on a dog, hers on an
antelope; he is clad in complete armour, the face and right hand
alone bare; the left hand holds the right-hand gauntlet, which he has
taken off before taking the lady’s hand. On the apex of the arch,
above the tomb, hangs the helm which he, during his life, used to
wear in tournaments. Their only child was born in 1441—Margaret,
of whom mention has been made. Her father died in 1444, aged
thirty-nine years, and the Duchy of Somerset became extinct in the
Beaufort family. His death took place at Kingston Lacy, an estate
close to Wimborne, belonging to the Beaufort family. His widow and
daughter went to live on the Bedfordshire property. In about four
years time, the widow married her third husband, Lord Welles. Young
Margaret, when only nine years of age, was sought by the Duke of
Suffolk as a wife for his son, John de la Pole, and by King Henry VI.
as wife for his half-brother, Edmund Tudor, Earl of Richmond, son of
Catherine, the Queen of Henry V., by her second husband, Sir Owen
Tudor. Margaret was a clever girl, well educated, knowing even then
Latin and French; but sorely distraught was she to know which of the
two suitors to choose: so she consulted an old gentlewoman, who
advised her to commit the matter to St. Nicholas. She took the
advice, prayed to the saint, and fell asleep, and about four o’clock
next morning, whether sleeping or waking she could not tell, saw one
standing in her room, habited in a bishop’s robes, who bade her
accept Edmund Tudor as a husband. She told her mother, and she
was betrothed to the Earl of Richmond, and they were married in
1455, when she was fourteen years of age and he twenty-four. They
lived at Pembroke Castle, which belonged to Jasper Tudor, Earl of
Pembroke. In 1456 her only son, Henry, afterwards Henry VII., was
born, and shortly after this her husband died. He was buried at
Caermarthen Abbey, and when the monastery was suppressed, his
body was removed to the Cathedral Church at St. David’s. His
mother, anxious to keep quite aloof from party strife (for the War of
the Roses had already broken out), lived on at Pembroke, educating
her son. In 1459 she married her second husband, Sir Humphrey
Stafford; widowhood, for one of exalted rank, not being a desirable
condition in those times of war and turmoil. It has been seen that her
mother was thrice married, and Margaret followed her example, for
when Sir Humphrey died in 1481, she, at the end of a year, being
then about forty years of age, married Thomas, Lord Stanley. After
fifteen years she separated from him with his consent, in order to
devote herself to a religious life, and retired to the convent at
Woking, in Surrey. It must not be supposed that she had a peaceful
or happy life. Her thoughts were centred on her only son, and many
were the years of separation from this son that his mother had to
endure. The story of his wanderings, his dangers, his detention in
Brittany, are too long to be told here—suffice it to say that Richard III.
became so odious to the chief nobles that at last it was arranged that
Henry Tudor, Earl of Richmond, Margaret’s son, should be
recognised by the Lancastrian party as their leader, and should claim
the throne; and that in order to gain the adherence of those
members of the Yorkist party who were opposed to Richard, a
marriage should take place between Henry and Elizabeth, the
daughter of Edward IV. Both the mothers agreed to the union: but
the first attempt at invasion by Richmond was a failure, and Richard
seemed free from all danger; and with a view to win over his Yorkist
opponents, he made up his mind to marry Elizabeth himself,
although she was his own brother’s child. This step led to a fresh
invasion—this time a successful one—and the death of Richard on
Bosworth field virtually placed the crown on Henry’s head (1485). So
at last the Lady Margaret’s troubles were at an end, and she saw her
son crowned and wedded, and the red and white roses twined
together. It remains only to speak of her benefactions. Those at
Cambridge are well known; and the writer of this chapter, who once
held at Christ’s College one of the scholarships she founded,
cherishes the memory of the royal and pious lady with all due
gratitude. But it is of her benefactions at Wimborne that mention
must now be made. About 1498 she built the beautiful monument to
her father and mother in the Minster; founded a chantry, where, for
her own soul, and for those of her son, her parents, and ancestors,
Mass was to be duly said. She founded and endowed the Grammar
School, though, as its endowments were greatly added to by Queen
Elizabeth, its name was afterwards changed from the Lady
Margaret’s to that of Queen Elizabeth’s Free Grammar School.
The Lady Margaret just outlived her son; both died in the same
year, 1509, but she rather later than he.
Of the other monuments space forbids mention of any, save that of
Anthony Etricke, if it can be called a monument, for it is really the
sarcophagus that contains the body of this eccentric magistrate, who
lived at Holt, and was recorder of Poole. He desired, for some
reason, to be buried neither above the ground nor under it, neither
within the church nor outside of it; and in order to carry out this
strange wish, he got permission to cut a niche in the south wall,
partly below the level of the surface of the churchyard, and in it fixed
a slate sarcophagus. In this he ordered his coffin to be deposited
when he died; and he made all the preparations he could
beforehand, even to the painting of the date of his death on the side
of the slate coffin. He had a presentiment that he should die in 1691,
but he lived till 1703, so that the real date of his death had to be
placed on the coffin. The other date was not obliterated, but the new
one was painted on the other, and the two may be seen there to-day.
The arms of his family are painted on the lid, and, as he left twenty
shillings per annum to keep coffin and niche in good repair, the
sarcophagus is bright and smart as paint, gilding, and varnish can
make it.
The Chained Library, Wimborne Minster.
The chained library, as it is called, is placed in a chamber above

the vestry, rebuilt when the church was restored. It was formed by
the Rev. William Stone, Principal of New Inn Hall, Oxford, afterwards
one of the “three vicars” of the church. By his will he left some land
to St. Margaret’s Hospital, and his collection of books to the Minster.
These were brought from Oxford in 1686, and placed in what was
then known as the Treasury. They were chiefly the writings of the
Fathers, with certain other theological books, and were intended for
the free use of the people of Wimborne. In these days few would
care to pore over these dry and heavy tomes, or make use of the
library, for it does not contain any novels, standard or ephemeral; but
there were days when it was used for study. In 1725 a catalogue was
made, and the number of the books was then two hundred; when
next catalogued, about one hundred and fifty years later, the number
had dropped to one hundred and eighty-five, despite the fact that ten
books not mentioned in the former catalogue were on the shelves.
There is but one MS., bearing date 1343, “Regimen Animarum,”
written on vellum, and containing some illuminated initials. The
majority of the books were printed between 1520 and 1710. The
most interesting are: a Polyglot Bible (1657), a Breeches’ Bible
(1595), and Sir Walter Ralegh’s History of the World (1614). No less
than one hundred and four pages of this History have had a hole
burnt through them. Tradition says that Matthew Prior, the poet, was
reading this book by candle light, and fell asleep; when he awoke he
found that some snuff from his candle had fallen on the book and
done the mischief. He stuck small pieces of parchment over the hole
in every page, and inserted the missing letters or words with pen and
ink. But the interesting tradition has been controverted in modern
times.
The books were originally chained to the shelves, one end of the
chain being fastened to the edge of the binding, the other to a ring
which would slide along a rod. Many of the books had got loose in
the course of years, and lay dusty and uncared for; but at the
restoration, new rods were fastened along the new shelves, and the
old chains repaired and put to their former use.
In two oaken chests in this room many deeds relating to the
Collegiate Church (the earliest dating from the time of Henry III., the
latest from that of Henry VIII.) are preserved. Among them is the
deed founding the Chantry and Grammar School, drawn up by the
executors of the Lady Margaret’s will; and also the charter granted
by Charles I. to the Governors of the church, from which they derive
the powers of appointing clergy, choristers, clerk, vergers, etc., which
powers they use to-day.
There is one other ancient ecclesiastical foundation in the
neighbourhood of Wimborne, about a quarter of a mile from the town
on the road to Blandford. It is now an almshouse, where three poor
married couples, three poor single men, and the same number of
unmarried women, are maintained; but its original purpose was to
relieve only such of the poor as were suffering from leprosy. It is
generally said to have been founded by John of Gaunt, and so to
have been another connection between that family and Wimborne.
There is, it may be said, an old kitchen at Canford which is still called
“John of Gaunt’s Kitchen”; whether he had anything to do with the
building of the kitchen or the endowment of the Lazar-house we do
not know, but it is certain that he did not found the latter, for, in the
reign of King John, Hugo of Lingiveria gave to it an acre of land, and
in 1282 the Bishop of Exeter gave an indulgence to any who would
contribute to its support. A deed of the date of Henry VIII. refers to a
Bull of Innocent IV., dated 1245, in which this hospital is mentioned.
Various gifts of land, vestments, plate, etc., were bestowed on the
hospital, to which a small chapel dedicated to St. Margaret and St.
Anthony is attached. A chantry was founded here by one John
Redcoddes, in order that a priest might daily say masses for his
soul’s welfare. The chapel, the architecture of which shows that it
was originally built in the thirteenth century, still stands, and is fitted
up for service. Hither once a week one of the clergy comes from the
Minster to conduct a service, which the almshouse people attend.
Other than the buildings already mentioned, there is little
mediæval work to be seen in Wimborne. The old Free Grammar
School buildings have given place to modern ones erected in 1851,
and the school is now managed by a governing body appointed
under a scheme drawn up by the Charity Commissioners. So “the
old order changeth, giving place to the new”; but, seen from far or
near, the two-towered Minster, with its parti-coloured walls of deep
red and drab stone, rises grand and old amid its modern
surroundings—a noble memorial of the mediæval builder’s art.
FORD ABBEY
By Sidney Heath
ARIOUS authorities agree with Camden in stating that
Ford Abbey (originally in Devon, but now included in the
county of Dorset), near Chard, was founded in the year
1140, for Cistercian monks, by Adeliza, daughter of
Baldwin de Brioniis, and a grand-niece of William the
Conqueror. The circumstances of its origin are interesting and
romantic. It appears that Adeliza’s brother, Richard of Okehampton,
had given, in 1133, certain lands at Brightley, within his barony, to an
Abbey of the Cistercian Order, and had secured twelve monks to
dwell therein from Gilbert, Abbot of Waverley, in Surrey. This small
community remained at Brightley for five years, when they, “by
reason of great want and barrenness, could abide there no longer,”
and commenced a return journey to their original home in Surrey. On
their way they passed through Thorncombe, the parish wherein Ford
is situated, where they encountered Adeliza, who, hearing with great
regret of the failure of her brother’s enterprise, exclaimed: “Behold
my manor where you now are, which is very fruitful and well wooded,
which I give you for ever in exchange for your barren lands at
Brightley, together with the mansion-house and other houses. Stay
there until a more convenient monastery may be built for you upon
some other part of the estate.” The site selected by the monks for
the erection of the Abbey was in a valley, on the left bank of the river
Axe, at a place called, according to Leland, “Hertbath” (balneum
cervorum), and which, from its nearness to a ford crossing the river
at this spot, subsequently became known as Ford.
Such is the accepted origin of the splendid pile of buildings which
sprang up in this fertile and sequestered valley in 1148, and which
still, notwithstanding the pillage at its dissolution, and its many
structural alterations, commands our admiration and our attention;
although, if we except some small portion of what is known as “the
chapel,” at the eastern end of the south front, nothing now remains
of the original foundation erected by the pious Adeliza.
The original purpose of this ancient part of the building, known as
“the chapel,” is somewhat obscure. It has been commonly regarded
as that portion of the religious house which its name indicates, and
as being the burial-place of its founder and other benefactors. Dr.
Oliver, however, in the supplement to his Monasticon, speaks of it as
the “Chapter House”—a likely suggestion. In his Memoir of Thomas
Chard, D.D., Dr. J. H. Pring writes:
That except in the deed of surrender, and a short
reference made to it by Hearne, I have not been able to
discover the slightest notice of “the Church of the Blessed
Virgin Mary of Ford” in any of the numerous accounts
which have been given of the abbey; though when we
read of frequent interments, some on the north, others on
the south side of the choir—others, such as that of Robert
Courtenay, who, we are told, was buried on the 28th July,
1242, in the chancel, before the high altar, under a stately
monument exhibiting the figure of an armed knight—there
can be little doubt, I think, that these took place, not in
what is now known as the Chapel, but in the Abbey
Church, which stood at the east end of the abbey, about
two hundred feet above the chapel.
Ford Abbey.
This portion of the edifice, whose original uses are conjectural,

shows, both inside and out, considerable vestiges which appear to
suggest a Norman origin, and which we may assume were possibly
erected under the immediate auspices, if not under the personal
superintendence, of the Lady Adeliza. The exterior angles of the
eastern end exhibit the quoins so characteristic of the Norman style
of building, and the interior has many fine examples of Anglo-
Norman work, in the pillars, the groined stone roof, the arches at
either end, of a slightly pointed character, with the well-known zig-
zag or chevron moulding. The eastern window is of much later date,
being Perpendicular in style, and it is believed to have been inserted
by Thomas Chard, the last Abbot, as the upper panel of the left-hand
side depicts a stag’s head, whilst the companion panel, parallel to it,
contains faint traces of the oft-repeated monogram, T. C.
The next feature in point of antiquity is what is now termed the
“Monks’ Walk,” a range of ivy-clad buildings running back for nearly
four hundred feet from the eastern end of the Abbey in a northerly
direction, and it is thought that a similar range ran parallel to it. The
remaining wing is on the eastern side, and consists of two storeys,
the lower of which possesses some beautiful Early English work, and
the upper one was probably the monks’ dormitory. In the centre is an
archway of fourteenth century date, and along the entire length of
the wing is a series of lancet windows, almost perfect on the western
side, but destroyed or built up on the eastern. Hearne thus notices
this wing:
But now, though one of the chief uses of the cloisters
was for walking, yet in Religious Houses they had
sometime galleries for the same end. We have an
instance of it in Ford Abbey in Devonshire, which is one of
the most entire abbeys in England; in the east front
whereof, which is the oldest of the two fronts (though the
south front be the chiefest), there is a gallery called the
Monks’ Walk, with small cells on the right hand, and little
narrow windows on the left.
Great as is the antiquarian interest of these fragments of what we
may reasonably presume to have formed part of the original
foundation, the greater part of the existing fabric is the work of Abbot
Chard, of whom we shall have something to say later. The best view
of the building is obtained from the front, where nearly all that meets
the eye affords a striking instance of the consummate taste and
devoted perseverance of this remarkable man under circumstances
that may well have discouraged the boldest. The storm which
culminated in the dissolution of the monastic houses was gathering;
but instead of being filled with dismay, as were so many of his fellow-
churchmen, Thomas Chard spared no effort to beautify his beloved
abbey, perhaps that the very glamour of her loveliness might
enchant the eyes of the spoilers and turn them from their purpose of
ruthless spoliation. To a great extent, his work was preserved, for,
although the abbey did suffer, and that grievously, yet it escaped the
wanton wreckage by which most of these foundations throughout the
land were devastated.
Sidney Heath 1907
Details from Cloisters. Ford Abbey.
The first portion of Chard’s building to claim attention is the

cloister, late Perpendicular in style, with mullions and window tracery
which present an appearance at once good and bold, and show no
signs of the debasement and formality that are so characteristic of
the late buildings of this period. Above the windows a frieze of
stonework depicts on shields the arms of various benefactors to the
Abbey—as those of Courtenay quartering Rivers, Poulett, the Bishop
of Exeter, etc.; and on many shields appear either the monogram or
the name of Thomas Chard.
An excellent account of the cloister—and, indeed, of the whole
Abbey—is contained in a very rare little volume, entitled, a History of
Ford Abbey, written anonymously many years ago, but
acknowledged by ecclesiologists to be the work of one who for a
long period must have resided there, and who thus, by daily
associations with the fabric, became more familiar with its minute
architectural details than could possibly be the case with anyone
who had not enjoyed a similar privilege. As this volume is rare, as
well as interesting and accurate in regard to its architectural
information, no apology is needed for quoting certain passages from
it here. In reference to the cloister we learn that:
The cloister is divided by a suite of rooms and arcade
from the grand porch-tower, so conspicuous for its
architectural beauty, and which in days gone by was no
doubt the original entrance. It is richly ornamented with
first-rate sculpture, some of it obviously unfinished; the
central boss in the vaulting uncut; and the blank shield in
the centre, below the basement window, encircled by the
garter, was doubtless intended for the royal arms. The
uncut shield on the sinister side, having the pelican and
dolphin for supporters, was for Courtenay. The two small
shields cut are charged with a lion rampant for De
Redvers, and cheeky two bars for Baldwin de Brioniis.
Immediately over the arch of the door is a large scroll
shield of a more modern date, bearing the arms of
Prideaux, impaling those of his second wife, Ivery. On the
upper part of this elegant specimen of Dr. Chard’s taste, in
the centre shield, are his initials, T.C., with the crosier and
mitre (Dr. Chard was a Suffragan Bishop); and the two
smaller shields, with the T.C., crosier, and abbot’s cap,
alternate with the stag’s head cabossed—supposed to be
the bearing of the then Bishop of Exeter; and just below
the battlement of the tower is the following inscription:—
AN̄ ’O D’N̄ I MILLESIMO QUINGESIMO VICMO OCTAO.

A D’N̄ O
FACTUM EST THOMA CHARD, ABB.
Now, while there is no doubt that Chard united in his own person
the offices of Abbot and Suffragan Bishop, the above account is at
fault in attributing “the stag’s head cabossed” to the then Bishop of
Exeter, for it formed no part of the armorial bearings either of Bishop
Oldham or of his successor, Veysey. In a letter from Dr. Chard to
Cardinal Wolsey “the stag’s head cabossed” is used as the seal, and
is expressly referred to in the body of the letter as “sigillum meum,”
and we find the same device associated with his name or monogram
in various parts of the Abbey buildings; the most probable solution
being that it relates to the ancient cognizance of the Abbey, or the
site whereon it stands, which, as we have already seen, was
Hertbath (balneum cervorum).
Panel from Cloisters. Ford Abbey.
Further confirmation of Dr. Chard’s double office of Bishop and

Abbot is found in a remarkable panel in the frieze (see illustration),
which appears to have been designed for the purpose of attesting
this fact, if not in actual words, yet in unmistakable and appropriate
symbolism. The small top corner shields of this panel contain the
letters T. C., and the lower ones an abbot’s and a bishop’s staff,
respectively; whilst on the hatchment-shaped panel in the centre
occurs the stag’s head and bishop’s staff, the name “Tho. Chard” on
a scroll entwined round an abbot’s staff; and above these, as a fitting
termination to the whole, appears the abbot’s cap, surmounted by
the bishop’s mitre.
The Chapel, Ford Abbey.
The entrance porch contains a fine west window of the same

character as those of the adjoining great hall, which in their turn
correspond with those of the cloister, and above them is a frieze of
grotesque animals. To quote once more from the book already
referred to:
This part of the building has been shorn of its length, as,
on minute inspection, will appear. The royal arms are not
in the centre, as they no doubt originally were. They
consist of a rose crowned, encircled with a garter, and
supported by a dragon and greyhound, the badges of
Henry VII.... Although the remaining portion of this wing
has been altered, it was built by Thomas Chard, the
battlements corresponding with the tower and chapel; and
as a more decisive proof that it was so, there is, at the
western end of the building, but hid by ivy, the portcullis
cut in stone, another of the badges of Henry VII.; and to
the north, or back side, are the initials T. C., with the
crosier and cap.
The ancient guest-chamber, so integral a part of these old
foundations, appears to have been at right angles to the great hall,
as it was noticed some years ago on the collapse of portions of the
ceiling that the ancient timber roof was still in situ. We shall have a
little to say later about the alteration and adaptation of the interior for
the purposes of a modern mansion, when, happily, much of Dr.
Chard’s work was not disturbed; but we have, unfortunately, no
record of the condition of the fabric prior to the restorations of the
above prelate, and his task seems to have been little less than the
rebuilding of the greater part of the edifice. The antiquary Leland,
visiting the Abbey during Dr. Chard’s alterations, writes: “Cœnobium
nunc sumptibus plane non credendis abbas magnificentissime
restaurat.”[47] This beautiful structure had scarcely had its delicate
stonework mellowed by the soft winds from the Devonshire moors,
when the Dissolution, long impending, burst in fury upon the larger
religious houses, and on March 8th, 1539, Thomas Chard was
induced to sign the surrender of his beloved Abbey of Ford, which
was endeared to him by many sacred associations, and on which he
had lavished his own private fortune and the artistic genius of a
master mind. The following is a translation (according to Dr. Pring) of
the document of surrender, the wording of which, we may be sure,
accorded ill with the reluctant hands that attached the names and
seals:—
To all the faithful in Christ, to whom this present writing
shall come: Thomas Chard, abbot of the monastery or
abbey, and of the Church of the Blessed Virgin Mary, of
Ford, in the county of Devon, of the Cistercian order, and
the same place and convent, everlasting salvation in the
Lord.
Know ye that we, the aforesaid
Per me Thomā abbem abbot and convent, by our
Willūs Rede, prior unanimous assent and consent, with
John Cosen our deliberate minds, right,
Robte Yetminster. knowledge, and mere motion, from
Johēs Newman. certain just and reasonable causes
r
Johēs Bridgwat . especially moving our minds and
Thomas Stafford. consciences have freely, and of our
Johēs Ffawell.
W. Winsor.
own accord given and granted, and
Elizeus Oliscomb. by these presents do give, grant,
William Keynston. and surrender and confirm to our
William Dynyngton. illustrious prince, Henry VIII., by the
Richard Kingesbury. grace of God, king of England, lord
of Ireland, supreme head of the
Church of England in this land, all
our said monastery or abbacy of Ford aforesaid. And also
all and singular manors, lordships, messuages, etc. In
testimony whereof, we, the aforesaid abbot and convent,
have caused our common seal to be affixed to these
presents. Given at our Chapter House of Ford aforesaid,
on the 8th day of the month of March, and in the thirtieth
year of the reign of King Henry aforesaid. Before me,
William Petre, one of the clerks, etc., the day and year
above written.
By me, Willm̄ n Petre.
No sooner had the document been signed than the work of pillage
commenced; but one is inclined to agree with the Devonshire
historian Prince, that, “by what lucky chance he knew not, Ford
Abbey escaped better than its fellows, and continueth for the
greatest part standing to this day.” At the same time, there is little
doubt that much havoc took place, although, perhaps, not to the
extent recorded by Risdon, who says it now merely “somewhat
showeth of what magnificence once it was.”
It is just possible that Thomas Chard’s beautiful work softened the
hearts of the spoilers, and its very wealth of ornament caused it to be
retained as too valuable a prize to be utterly demolished; but,
whether standing entire or razed to the ground, it appears to have
been an encumbrance, for on October 28th, in the year of its
surrender, it was granted by the King, “with all and singular its
manors, lordships, and messuages, etc.,” to Richard Pollard, Esq.
At the time of its dissolution the annual revenues of the Abbey
were computed at £374 10s. 6¼d. by Dugdale, and at £381 10s. 6d.
by Speed, and the net revenue was, no doubt, somewhere between
these two sums.
Born probably at Tracy, near Awliscombe, Honiton, about the year
1470, Thomas Chard was one of the most distinguished
ecclesiastics of his day, and evidently, as his works attest, an
accomplished architect and a most munificent man. The highly
ornamental façade of the institution over which he presided as last
abbot is considered to be the finest example of its kind in the West of
England. On entering holy orders, Chard appears to have held
several livings in Somerset, Devon, and Cornwall, and was elected
Abbot of Ford about 1520. Previous to this, in 1508, he was
appointed Suffragan to Bishop Oldham by the title “Episcopus
Solubricencis,” in 1513 Warden of the College of Lady St. Mary, at
Ottery, and in 1515 Prior of the Benedictine or Cluniac Priory of
Montacute. It has been suggested that as Dr. Chard was Warden of
Ottery College about the time that the beautiful Dorset Chapel was
built (1513-18)—one of the most lovely pieces of Perpendicular
building we possess—the inspiration of this eminent architect may
have done much to influence the splendid design of this portion of
the Church of Lady St. Mary at Ottery.
It was Bishop Chard who officiated for Bishop Veysey, of Exeter, at
the noble obsequies of Katherine Courtenay, daughter of Edward IV.,
and widow of William Courtenay, Earl of Devon, buried at Tiverton in
1527. It is thought that his choice for this office was determined by
his headship of the Monastery of Ford, of which foundation the
Courtenays had always been great patrons and benefactors.
The burial place of Thomas Chard is unknown, but may possibly
be in the chapel of the Hospital of St. Margaret, near Honiton. Dr.
Oliver, who visited this chapel many years ago, writes: “The west
door is secured by a large sepulchral slab, to which was formerly
affixed a brass plate.”
This has long since
disappeared, but many
writers agree that there
is little doubt that this
slab covered the dust of
the Abbot-Bishop.
The old abbey
seal,[48] which had
eluded the research of
many antiquaries,
including the editors of
Dugdale’s Monasticon,
was discovered by Mr.
Davidson, of Sector,
near Axminster. It is of
oval form, the usual
shape for monastic
seals, and is divided
into three
compartments, in the
Seal of Ford Abbey.
Full Size
uppermost of which is a
bell suspended in a
steeple, and in the
canopy beneath we see the Blessed Virgin with the Divine Infant on
her knee. On one side is the shield of Courtenay, bearing—or, three
torteaux, with a label of three points. On the other side is the shield
of Beaumont—barry of six, vair and gules. The lowest compartment
occupies rather more than half the seal inside the inscription, and
shows an abbot standing, in his right hand a pastoral staff, and
holding in his left hand a book; and at his feet are three monks
kneeling, with their hands together in supplication.
With this description of the seal the claims of Ford Abbey to figure
in this volume of “Memorials” are practically finished, yet it may be of
interest to continue a little further in the personal and architectural
history of this wonderful old house. As we have seen, Henry VIII.

Buy Ebook Mathematical Problems in Meteorological Modelling 1st Edition András Bátkai Cheap Price

Uploaded by

Copyright:

Available Formats

Buy Ebook Mathematical Problems in Meteorological Modelling 1st Edition András Bátkai Cheap Price

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Buy Ebook Mathematical Problems in Meteorological Modelling 1st Edition András Bátkai Cheap Price

Uploaded by

Copyright:

Available Formats

Full download test bank at ebook textbookfull.

CLICK LINK TO DOWLOAD

Mathematical Modelling in Real Life Problems Case

Mathematical Modelling of Waves in Multi Scale

Mathematical Modelling in Solid Mechanics 1st Edition

Mathematical Modelling 1st Edition Seppo Pohjolainen

Modelling in Mathematical Programming: Methodology and

Language and Culture in Mathematical Cognition Volume 4

Exercises and Problems in Mathematical Methods of

Methods of Mathematical Modelling: Fractional

THE EUROPEAN CONSORTIUM

István Faragó András Horányi

ISSN 1612-3956 ISSN 2198-3283 (electronic)

Library of Congress Control Number: 2016954078

Mathematics Subject Classification (2010): 86A10, 65L06, 62H11

© Springer International Publishing Switzerland 2016

Printed on acid-free paper

This Springer imprint is published by Springer Nature

De Bilt, Netherlands Jan Barkmeijer

Meteorological modelling, and in particular numerical weather prediction, belongs

Budapest, Hungary András Bátkai

Part I Numerical Methods for Meteorological Problems

3 Turbulence Modeling Using Fractional Derivatives . . . . . . . . . . . . . . . . . . . . 47

Part II Air Quality Modelling

6.5 Numerical Method for Solving the Non-stationary

Part III Meteorological Data Assimilation and Probabilistic

11 Ensemble Methods in Meteorological Modelling . . .. . . . . . . . . . . . . . . . . . . . 207

The dynamics of the atmosphere is governed by physical, chemical, and even

Alexander Zlotnik and Vladimir Gavrilin

Keywords 1D shallow water equations • Discrete energy balance law • Parabolic

© Springer International Publishing Switzerland 2016 3

1.2 The 1d Shallow Water System of Equations,

@t .hu/ C @x .hu2 / C @x p D @x ˘NS C hF; (1.2)

p D 0:5gh2 ; ˘NS D @x u; F.x; t/ D g@x b.x/ C f .x; t/; (1.3)

where g D const > 0, D .h/ > 0 is the viscosity coefficient (when D 0,

h.x; 0/ D h0 .x/; u.x; 0/ D u0 .x/; 0 6 x 6 X: (1.4)

The regularized 1d shallow water system of equations consists of the modified

@t h C @x .h.u w// D 0; (1.5)

@t .hu/ C @x .h.u w/u C p/ D @x ˘ C h F: (1.6)

For f D 0 and suitable boundary conditions at x D 0; X, for example, for the

@x p.hS / C ghS @x b D 0 on .0; X/; (1.12)

i.e., simply to the formula hS .x/ C b.x/ C on Œ0; X.

ı.uv/ D ıu Œv C Œuıv; (1.13)

ı .yŒv/ D ı y v C Œ yıv (1.14)

(u is given on !N h ). To reduce the amount of brackets, hereafter we suppose that,

Turning to [23], we first construct a three-point symmetric in space discretization

where B WD 0:252C .ıp C gh ıb/ıu.

according to formula (1.14), we get

Next we multiply the momentum balance equation (1.18) by u. We apply the

@t .hu/ u D 0:5@t .hu2 / C 0:5@t h u2 ;

Then also taking into account elementary formulas

Œu2 D 0:5Œu2 C 0:5u uC ; 0:5ı.u2 / D Œuıu (1.24)

and formula (1.15), we derive

0:5@t .hu2 / C 0:5ı . ju uC / C Œıp C gh ıb u ı ˘ u D Œh f u:

with the above introduced B . Since ıp C gŒhıb D Œhgı.h C b/, we obtain

with A WD j fg.h C b/ C 0:5u uC g ˘ Œu.

 .h; u/ WD Œhgı.h C b/.w C f / Œuı.hu/ gıb

D Œhgı.h C b/.w C f / C ŒhŒuıu .wO C f /

gıb Œuı.hu/ C gŒhı.hu/ ıu:

where B WD 0:252C .ıp C gh ıb/ıu.

with the above introduced B . Since ıp C gŒhıb D Œhgı.h C b/, we obtain

.h; u/ WD Œhgı.h C b/.w C f / Œuı.hu/ gıb

see (1.21), we transform .h; u/ as follows