Original scientific paper
THE MUST MODEL EVALUATION EXERCISE: PATTERNS IN MODEL PERFORMANCE
H.R. Olesen
; NERI, Univ. of Aarhus, Denmark
A. Baklanov
; DMI, Denmark
J. Bartzis
; Univ. of West Macedonia, Greece
F. Barmpas
; Aristotle Univ., Greece
R. Berkowicz
; NERI, Univ. of Aarhus, Denmark
K. Brzozowski
; Univ. of Bielsko-Biala, Poland
R. Buccolieri
; Univ. of Salento, Italy
B. Carissimo
; CEREA, France
A. Costa
; Univ. of Aveiro, Portugal
S. Di Sabatino
; Univ. of Salento, Italy
G. Efthimiou
; Univ. of West Macedonia, Greece
J. Franke
; Univ. of Siegen, Germany
I. Goricsan
; Budapest Univ. of Technology and Economics, Hungary
A. Hellsten
; Helsinki Univ. of Technology, Finland
M. Ketzel
; NERI, Univ. of Aarhus, Denmark
B. Leitl
; Univ. of Hamburg, Germany
R. Nuterman
; Tomsk State Univ., Russia
E. Polreich
; ZAMG, Austria
J. Santiago
; CIEMAT, Spain
R. Tavares
; Univ. of Aveiro, Portugal
Abstract
As part of the COST 732 action more than a dozen different research groups have modelled the MUST experiment, as
simulated in a wind tunnel. The model evaluation guidance developed within COST 732 recommends 'exploratory data analysis' as
one of the elements in model validation. Experience has shown that such exploratory analysis is crucial to reveal shortcomings of
models that might otherwise pass unnoticed. Conditions are best for detecting common patterns and anomalies if you have a
situation where several models are put into a common framework – like the case at hand. The available material provides a unique
opportunity to identify and explore patterns within model performance.
Keywords
COST 732,MUST; CFD; atmospheric dispersion; model evaluation; model validation; exploratory data analysis
Hrčak ID:
64279
URI
Publication date:
12.12.2008.
Visits: 812 *