This site http://www.airinfonow.org/html/ed_ozone.html contains the following description of  "Things That Affect Ozone Levels"

"The amount of ingredient gases.

More ingredient gases (VOC's and NO2) can mean more ozone. Towns and cities that have more traffic or more industrial plants have a higher potential for ozone formation, especially towns that also experience many warm sunny days with little wind.

Weather and Seasons: warm sunny days mean more ozone.

Since sunlight and heat are the engines that drive ozone formation, warm sunny days should have more ozone than cool or cloudy days. Wind can also play a role. On windy days the wind can disperse the ozone, causing levels to drop. Ozone pollution can be especially bad during summer heat waves when the air does not mix very well and air pollution doesn't disperse.

Time of Day: ozone levels build during the day.

On a clear day, ozone levels can continue to rise all day long, and then decrease rapidly after sunset. Since heat, sunlight, the ingredient gasses each usually increase during the day, ozone formation also increases. When the sun goes down, there is no energy for ozone formation and fewer ingredient gases - so ozone levels drop.

From that it would seem unlikely that you would record high levels of ozone in the winter.

Ozone formation is depended on sunlight, high latitude mean a lower angle, shorter daytime and higher number of cloudy days in winter. Typically, NOx concentrations are increasing during winter time (heating as example) and biogenic VOC are low (plants are not/less active) that changes ozone formation during winter conditions. Low temperatures play also a role in determining the speed of processes such as formation and growth of aerosol particles. Low temperature mean lower reaction speed. I guess, that the chemistry in your model follow these principles and ozone impact on air quality is therefore lower in winter than during a summer scenario.

Thanks for answers but I was talking about chemical  data assimilation performance of ozone and its seasonal dependance (summer vs winter). Not the seasonal summer vs winter concentrations of ozone produced by the model which is a very well known topic as you have described. Although low concentrations of ozone in winter could be one explanation, I would appreciate to know if anybody has performed a comparative study of  surface chemical data assimilation versus season (summer vs winter)  for surface ozone and what was the results. Or maybe nobody bothered to do that.

What about straosph. intrusion (the largest source for ozone in my country in winter) and data assimilation?

Wintertime ozone may be simply occurring due to vertical fumigation occurring from free troposphere where precursors may react to generate Ozone under solar irradiation. Ground levels in wintertime is low because of the more pronounced atmospheric stability. However, especially during sunny days, midday turbulence may bring some free troposphere Ozone down in the ground.

I am still not clear what you are asking, and if I had any sense I would shut up :-)

But I found this paper which seems to be saying the same as you:

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 112, D12S21, doi:10.1029/2006JD007722, 2007

at:

https://www.researchgate.net/profile/Thomas_Fairlie/publication/228745253_Chemical_data_assimilation_estimates_of_continental_US_ozone_and_nitrogen_budgets_during_the_Intercontinental_Chemical_Transport_ExperimentNorth_America/links/0fcfd5111657ba411d000000.pdf

and this web page which may provide you with some leads:

http://www.ukca.ac.uk/wiki/index.php/UKCA

Hope this helps.

Article Chemical data assimilation estimates of continental US ozone...

Ok, now I understand your question better. I can't give you a full conclusive answer but maybe some hints. Numerical inversion methods have to assume some process or condition to be valid. It may happen, that this assumption does not hold under certain conditions. An example is the assumption that model and observational errors follow a Gaussian distribution, that must not hold always. The assumption of a well mixed atmosphere must not hold if comparing day and night situations as another example. If in wintertime the data are only weakly reflecting the underlying assumption or process, then the benefit of the data assimilation will be also weak or neglectable.