Comment on “Enhanced upper stratospheric ozone: Sign of recovery or solar cycle effect?” by W. Steinbrecht et al.

Abstract

[01] Steinbrecht et al. [2004] (hereinafter referred to as S4) have discussed the trend in upper stratospheric ozone at 35–45-km altitude determined from their lidar measurements at Hohenpeissenberg (47.8 N, 11.0 E) from 1987 to 2003. They question the conclusion of Newchurch et al. [2003] (hereinafter referred to as N3) that after approximately 1997 the downward trend of upper stratospheric ozone at 35–45-km altitude has diminished significantly. They argue instead that recent ozone changes are associated with the recent solar maximum (i.e., the solar cycle effect on ozone). In this comment we question their procedure for identifying the solar cycle effect. Moreover, we argue that the solar cycle effect was appropriately accounted for in the N3 analysis, and we buttress our argument by demonstrating that the more extensive data set used by N3 shows that the trend in upper stratospheric ozone has diminished significantly since 1997 and that this is evidence of the first stage of ozone recovery. [2] In the following text, unless specifically indicated otherwise, we will be comparing and contrasting here, as Steinbrecht et al. [2004] did, the Hohenpeissenberg analysis at 35–45-km altitude against the Stratospheric Aerosol and Gas Experiment (SAGE)-Halogen Occultation Experiment (HALOE) analysis at 35–45-km altitude and 30 –50 N. We do note that the single-station lidar record is remarkably representative of the longer global SAGE and HALOE records during the time period of coincidence (1987– 2003). Newchurch et al. [2003] used the 10.7-cm solar radio flux (F10.7cm flux) as a proxy for the temporal variability of the effect of the solar cycle on upper stratospheric ozone. S4, on the other hand, inferred the effect of the solar cycle using a multiple linear regression approach based on a combination of 10-, 13-, 67-, and 135-month periodicities derived from 193 months of ozone lidar data. In each case, the amplitude of the solar cycle effect was obtained by fitting the F10.7cm flux (N3) or the harmonics (S4) to the ozone observations, resulting in a peak-to-peak solar effect of approximately 3% given by N3 and 7% given by S4. It should be noted that S4 commented that when they repeated their calculations using the F10.7cm flux as the proxy instead of using harmonics they obtained a solar cycle effect in the differential absorption lidar (DIAL) observations virtually the same as the result that N3 obtained from SAGE. Therefore S4’s use of harmonics as the proxy for representing the solar cycle effect and the longer period of data used by N3 (1979–2003) are the two major factors producing the different conclusions obtained by N3 and S4. [3] The F10.7cm flux or the Mg II line emission constitute the standard procedures used by both modelers and analysts for representing the solar cycle effect because both parameters are proxies for variations in the ultraviolet flux that cause ozone change [e.g.,McCormack and Hood, 1996; Lee and Smith, 2003]. We also favor the solar flux proxy approach because it is physically based. Nevertheless, in order to provide a relatively direct comparison between the N3 and the S4 results we have calculated the best fit harmonics for the SAGE data over the lidar period from 1987 to 2003 (after having first removed a small mean difference between the two sets of measurements and a quasi-biennial oscillation (QBO), i.e., following the same initial steps used by both N3 and S4). This optimal harmonic fitting procedure results in 127and 63-month components over the full SAGE/solar backscatter ultraviolet instrument (SBUV) period from 1979 to 2003; 140and 70-month components from the SAGE data over the shorter lidar period 1987–2003; and 135and 67-month components from the lidar data fitted by S4. Figure 1 shows a comparison of these solar cycle fits: SAGE/SBUV full period, 127and 63-month components, dashed black trace; SAGE optimal fit over the lidar period, 1987–2003, 140and 70-month components, dotted red trace; and SAGE 1987–2003 fit with the S4 lidar-fitted components of 135 and 67 months, solid black trace. JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 109, D14305, doi:10.1029/2004JD004826, 2004