Application of a multiple regression (MR) statistical model to the TOMS/SBUV column ozone data set calibrated at Goddard Space Flight Center yields evidence for a significant dependence on latitude, longitude, and season of the solar regression coefficients. First, regression coefficients are generally larger at subtropical to middle latitudes and are smallest near the equator for all seasons. In NH winter, the responses are especially large in the northern subtropics and are stronger in some longitude sectors than in others. In particular, an especially strong response maximum (4 to 5%) is obtained in the North Pacific centered at 30 N while the response is relatively weak (< 2%) in the eastern equatorial Pacific. Stronger responses are also obtained in the SH subtropics. The NH winter solar response is similar in some respects to the NH winter ENSO response. Specifically, the ENSO winter column ozone response is negative at low latitudes, especially in the eastern Pacific but is positive at some longitudes at subtropical to middle latitudes (e.g., Hood et al., JGR, 2010). Tests using an MR model with and without an ENSO term show that aliasing between the ENSO and solar responses is minor and cannot explain the observed regional responses to 11-year solar forcing. Since the ENSO response is driven entirely by internal ocean-troposphere variability, these results strongly suggest that coupling to the ocean-troposphere system is involved in producing the lower stratospheric solar cycle response.
To investigate further the origin of the pronounced regional signals in the NH winter column ozone response to solar forcing, we have carried out some preliminary comparisons with the ozone response calculated using a chemistry climate model that includes a coupled troposphere and ocean. Specifically, we consider the simulations reported by Meehl et al. (Science, 2009) using the WACCM3 CCM with and without coupling to the ocean, land, and sea ice components of the Community Climate System Model 3. The latter authors reported results that appear to provide an improved agreement with observations of solar cycle surface climate signals in the Pacific region. It should be noted that WACCM3 has recently been upgraded to WACCM4, which has been more carefully validated and also includes a coupled ocean (D. Marsh, private communication, 2010). Nevertheless, some initial comparisons with the WACCM3/CCSM3 simulations may be useful. The latter simulations did not include an imposed QBO; however, the MR technique effectively averages over both phases of the QBO so the comparison is still justified to first order.
We find that the WACCM3 simulations with coupling to CCSM3 generally provide a better agreement with the observed stratospheric ozone response in NH winter and spring than do the simulations without coupling to CCSM3. First, the coupled simulations produce annual mean ozone responses in the tropical and subtropical lower stratosphere that are stronger and more distributed in latitude than those of the uncoupled simulations. There is also some evidence for subtropical ozone response maxima in the lower stratosphere for the coupled simulations. In addition, the NH winter column ozone response obtained using the coupled simulation results has subtropical maxima in the Pacific sector that resemble those estimated from the satellite data. In particular, a positive regional model response is obtained in the North Pacific that is associated with a low model sea level pressure (SLP) anomaly in the same region. The SLP anomaly is likely due to higher model sea surface temperatures in the warm pool region of the western Pacific, which are known to strongly impact the response of models in the North Pacific (Deser and Phillips, J. of Climate, 2006). The low SLP anomaly is in the same region as the observed North Pacific column ozone response. Comparisons for NH spring also show an improved agreement of the coupled simulation column ozone responses with the observed responses. Stronger model ozone responses are obtained in the NH subtropics and middle latitudes that are associated with low SLP anomalies at the same latitudes. Overall, the results therefore support a significant role for feedbacks from the troposphere-ocean response in producing the lower stratospheric solar cycle responses during boreal winter and spring.18.104.22.168 on 7-27-2010-->