<p><strong>Abstract.</strong> Downward transport of ozone (O<sub>3</sub>) from the stratosphere can be a significant contributor to tropospheric O<sub>3</sub> background levels. However, this process often is not well represented in current regional models. In this study, we develop a seasonally and spatially varying potential vorticity (PV)-based function to parameterize upper tropospheric and/or lower stratospheric (UTLS) O<sub>3</sub> in a chemistry transport model. This dynamic O<sub>3</sub>–PV function is developed based on 21-year ozonesonde records from World Ozone and Ultraviolet Radiation Data Centre (WOUDC) with corresponding PV values from a 21-year Weather Research and Forecasting (WRF) simulation across the Northern Hemisphere from 1990 to 2010. The result suggests strong spatial and seasonal variations of O<sub>3</sub><span class="thinspace"></span>∕<span class="thinspace"></span>PV ratios which exhibits large values in the upper layers and in high-latitude regions, with highest values in spring and the lowest values in autumn over an annual cycle. The newly developed O<sub>3</sub><span class="thinspace"></span>∕<span class="thinspace"></span>PV function was then applied in the Community Multiscale Air Quality (CMAQ) model for an annual simulation of the year 2006. The simulated UTLS O<sub>3</sub> agrees much better with observations in both magnitude and seasonality after the implementation of the new parameterization. Considerable impacts on surface O<sub>3</sub> model performance were found in the comparison with observations from three observational networks, i.e., EMEP, CASTNET and WDCGG. With the new parameterization, the negative bias in spring is reduced from −20 to −15<span class="thinspace"></span>% in the reference case to −9 to −1<span class="thinspace"></span>%, while the positive bias in autumn is increased from 1 to 15<span class="thinspace"></span>% in the reference case to 5 to 22<span class="thinspace"></span>%. Therefore, the downward transport of O<sub>3</sub> from upper layers has large impacts on surface concentration and needs to be properly represented in regional models.</p>