Reply [to “Comment on ‘Tropospheric OH in a three‐dimensional chemical tracer model: An assessment based on observations of CH₃CCl₃’ by C. M. Spivakovsky et al.”]

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 96, NO. D9, PAGES 17,389-17,390, SEPTEMBER 20, 1991 Reply C.M. SPIVAKOVSKY, R. YEVICH. J.A. LOGAN. S.C. WOFSY, AND M.B. MCELROY Division of Applied Science and Department of Earth and Planetary Sciences, Harvard University, Cambridge M. J. PRATHER NASA Goddard Institute for Space Studies, New York concurrent data for CC12F2, CFCI•, Hartley and Prinn [this issue] misrepresent the objectives, pro- years of yet-to-be-published cedures, and results of our work [Spivakovsky et al., 1990, and CHzCC13. As emphasized in our paper, observations of hereinafter referred to as S90]. In addition, their analysis is longer-lived species (e.g., CC12Fa and CFClz) provide invaluable plagued by a number of conceptual errors and misuse of statistics. additional constraints on the relative influences of chemical and The objective of S90 was to compute the global three- dynamical processes in controlling the distribution of CHzCClz. dimensional field of tropospheric OH, based on our understanding Besides, the ancillary information tells a different tale. of atmospheric chemistry and the climatologies of precursors, and The observations at Samoa available to S90 provide evidence then to assess the results using available observations. This work for an annual cycle in CFCs which is satisfactorily reproduced by also included an extensive analysis of observations of CFCs and our model. Important differences were noted in S90 between the CH3CC13 in order to identify and compare annual cycles of these seasonal behavior of CFCs and CHzCC13 observed at Samoa. species. Comparisons of seasonal variations of chemically passive Observations of CH•CClz, unlike those for CFCs, did not display tracers (CFCs) with those for CH3CCI5, which is destroyed by a significant annual cycle. We suggested that the differences reaction with OH, were intended to distinguish annual cycles asso- could reflect cancellation (for CHzCCI•) between the chemical ciated with transport from those due to chemistry. We concluded and dynamical components of the variations which appeared to be that seasonal variations of CH3CC13 are dominated by chemistry of opposite phase and comparable amplitude (see Figure 24 in only at southern mid-latitudes and that only for this region can S90). Hartley and Prinn could have made a useful contribution in they provide constraints for OH. this context. Do the 11-year records for CFCs and CH3CClz sup- Hartley and Prinn seem to believe that the primary purpose of a port our suggestion? modeling study is to allow an adjustment of model sources and We were pleased to see that the 11-year record for CH•CCI• sinks either by trial and error or by an inverse method until the displays an annual cycle at Barbados similar to the one we showed model agrees with data. This was never the intent of our studies. for the published data. Apparently, our coarse-resolution global Data for most Uace species are inadequate to allow unambiguous model reproduces the magnitude and, approximately, the phase of determination of either sources or sinks, except on a hemispheric or global scale. Indeed, the purpose of S90 was to explore what could be learned about OH from analysis of the ALE/GAGE data. We concluded that competitive effects of transport and chemistry were such that concentrations of OH could be constrained only in a global-average sense with somewhat greater detail for southern mid-latitudes. In particular, the latitudinal distribution of OH can- not be constrained on the basis of data from the five ALE/GAGE the fall minimum. Our simulations suggested that this seasonal decrease is determined by a dynamical rather than a chemical sig- n fl. However, observations of CFCs at Barbados available to S90 do not display a significant annual cycle. Do the 11 years of observations of CFCs define a cycle at Barbados similar to that for CH•CC137 It is incorrect to expect a single model year (or, for that matter, a single year of observations) to fall within the standard error of stations. the mean of observations as implied by Hartley and Prinn. The Hartley and Prinn address their attention to a peripheral com- results shown in S90 do not represent averaged seasonal variations ponent of our paper, the month-to-month variability of CH•CC13 but rather typical seasonal variations, since the simulations were in the tropics. The resolution of the model employed in our study, based on a single year of GCM statistics. Hartley and Prinn state combined with use of a single year of dynamical output from the that the interannual variabilifies ... for Samoa are in fact multimo- general circulation model (GCM), precluded adequate simulation dal. How then should one interpret the standard deviations in of the movement of the intertropical convergence zone (1TCZ) or Figures 4a and 5a? What fraction of the observations is expected simulation of the interannual variability of tropical meteorology to fall within one standard deviation of the mean? Is there reason (related for example to E1 Nino Southem Oscillation (ENSO). to expect that the intra-annual variations simulated for Samoa The assumptions used in our simulations are clearly stated in S90. should agree with observations averaged over 11 years given that Hartley and Prinn ereate and demolish a straw man adding little to observed seasonal variations appear to be nonrecurrent, i.e., indis- our understanding of tropical meteorology. tinguishable from random (the appropriate autocorrelation coeffi- Regrettably, Hartley and Prinn restrict their discussion to the cient is equal to -0.1 as shown in Figure 3b of the comment)? use of data for CHa CCla, unfortunate in light of authors' access to Hartley and Prinn in their calculation of correlation coefficients Copyright 1991 by the American Geophysical Union. Paper number 91JD01670. 0148-0227/91/91JD-01670502.00 for the different ENSO phases appear to have treated the separated 12-month periods as continuous. This is likely to produce spuri- ous correlations or anticorrelations; consequently, results for ENSO warm events, ENSO cold events, and intermediate years are suspect (Figure 3b in the comment).