Zabkar, Rahela and Joze Rakovec, University of
Ljubljana and Center of Excellence SPACE‐SI; Darko Koracin,
Desert Research Institute, Reno
In the present work
results of an extensive tropospheric ozone sensitivity modelling study
performed using the WRF/Chem model as part of the research activities related
to the recently implemented operational ozone forecast in Slovenia, are
presented. The motivation came from results of a previous work focused on the
roles of regional vs. local emission sources on ozone levels in Slovenia.
During the high ozone episodes in Slovenia the main dynamic characteristics of
episodes were generally well simulated, but some significant discrepancies
between simulated and measured ozone and basic meteorological variables were
common to all analyzed high ozone episodes, including underestimation of
daytime ozone levels in Slovenian coastal areas and overestimation of nighttime
ozone levels at some stations.
The aim of the present
study is: (1) to investigate the previously observed model systematic errors in
simulated ozone levels, (2) to explore ability of different WRF physics
parameterization schemes to accurately simulate meteorological conditions
typical of high ozone episodes over a region of the geographically complex
northeastern Mediterranean Basin, and (3) to investigate the sensitivity
results of simulated ozone levels due to the main sources of model
uncertainties (meteorological boundary layer processes, anthropogenic and
biogenic emissions, lateral and initial chemical boundary conditions). To accomplish the study goals, more
than 50 experimental ensemble simulations were performed for an episode that
was highly favorable for ozone formation. The experimental runs consisted of
different WRF model physical parameterizations, including different planetary
boundary layer, land surface model, surface layer, radiation schemes, and with
improved sea surface temperature. In addition, some tests regarding other major
sources of model uncertainties (e.g., biogenic and anthropogenic emissions,
their temporal variations, and chemical boundary conditions) were also included
in the ensemble runs to examine the influence of these uncertainties on model
results.