Original scientific paper
USER EXPERIENCE WITH MODEL VALIDATION EXERCISES
Kathrin Baumann-Stanzer
; Central Institute for Meteorology and Geodynamics, Vienna, Austria
Martin Piringer
; Central Institute for Meteorology and Geodynamics, Vienna, Austria
Erwin Polreich
; Central Institute for Meteorology and Geodynamics, Vienna, Austria
Marcus Hirtl
; Central Institute for Meteorology and Geodynamics, Vienna, Austria
Erwin Petz
; Central Institute for Meteorology and Geodynamics, Vienna, Austria
Marianne Bügelmayer
; Central Institute for Meteorology and Geodynamics, Vienna, Austria
Abstract
Gaussian and Lagrangian model runs are evaluated in comparison to field data from the Odour Release and Odour
Dispersion project and to wind tunnel data from the Mock Urban Setting Test (MUST). Different statistical metrics are discussed.
To conclude which model performs best in the two cases, a weighted multiplier proposed by Sornette et al. (2007) is calculated
based on each metric and finally multiplied to one score per model and experiment. The results illustrate once again that a good
model performance is strongly dependent on the model input (e.g. terrain data, roughness length). Promising results are received
from a combination of the Lagrangian dispersion model LASAT with wind field simulations calculated with the CFD model
MISKAM.
Keywords
model validation; statistical measures; Gaussian and Lagrangian models; COST 732
Hrčak ID:
64157
URI
Publication date:
12.12.2008.
Visits: 1.734 *