cc: wigley@meeker.ucar.edu, boyle@pcmdi.llnl.gov, hnilo@pcmdi.llnl.gov, mwehner@pcmdi.llnl.gov, ktaylor@zooks.llnl.gov, doutriau@pcmdi.llnl.gov, bengtsson@dkrz.de, roeckner@dkrz.de, esch@dkrz.de, meehl@meeker.ucar.edu, p.jones@uea.ac.uk date: Tue, 07 Mar 2000 11:16:34 -0500 from: Dian Gaffen subject: Re: [Fwd: Technical Comment on RA1046025] to: Ben Santer Dear Ben and Everyone, It's too bad we have to deal with this. However, reading between the lines of MKS's paragraph 3 and their quotation of the original paper, it seems that they may feel we provoked them in footnote 37, in which case I suppose we bear some responsibility. It's interesting that they do not take on the differential trend issue, only the 2LT trend. Ben's synopsis of the issues seems on the mark to me. I'd suggest dealing with these in as straightforward a manner as possible, relying as much as possible on the literature, rather than further analysis. This would be easier for us and would bolster the arguments by bringing in other people's work to support our contentions. An upcoming paper by Jim Angell effectively deals with what Ben identifies as issue 2, the El Chichon effect. Without going into details, I quote the following relevant paragraph from Jim's paper. "Even though the amplitudes are very different, the 850-300 mb temperature variations for tropics and globe parallel the Nino3 SST variations well, except following volcanic eruptions. In particular, the tropospheric warming associated with the strong 1982-83 El Nino overwhelmed the cooling due to the El Chichon eruption in 1982, but the cooling due to the Pinatubo eruption in 1991 overwhelmed the warming associated with the modest 1991-92 El Nino. Note the tendency for the tropical 850-300 mb temperature variations to lag the Nino3 SST variations and, less obviously, the global 850-300 mb temperature variations to lag the tropical 850-300 mb temperature variations." If we want to quote this result directly, we don't need to include the final sentence. This issue of lag is taken up later in the paper; Jim finds maximum correlation at a lag of 2 seasons, for both the tropics and globe, although there is a suggestion of a slight difference that might be discernible if higher temporal resolution data were used. I include this as possibly applicable to analysis of the model simulation. Reference: Angell, J.K., Tropospheric temperature variations adjusted for El Nino, 1958-1998. J. Geophys. Res., in press. (Jim's done the galley-proofing, so this could appear soon.) FYI, a new manuscript by Melissa Free and Jim Angell confirms this result using the HadRT gridded CLIMAT TEMP data, whereas Jim's result is based on daily radiosonde data for his 63-station network. Melissa and Jim also deal with the vertical structure of the temperature response to Agung, El Chicon, and Pinatubo, which is quite relevant to our work. Their manuscript is still in NOAA internal review, however, so we should not cite it in our reponse. Assuming we can effectively argue that the trend increment proposed by MKS for El Chichon makes no sense given the larger El Nino effect then, we have only to deal with the ENSO increment, which I'll leave to others to address. But, even if we accept this increment as reasonable, we should point out that the MOD+ENSO is not significantly different from OBS in the MKS figure if the confidence interval on the OBS and MOD trends are taken into account. (For that matter, even the MOD+ENSO+VOL trend lies within the confidence interval of the OBS trend, if the values from Table 1 are used.) If the MKS piece gets published with this figure, we should insist that confidence intervals be plotted with the trends. I hope some of this is useful. What a way to spend Mardi Gras and Super Tuesday. Laissez les bon temps roulez! Dian Ben Santer wrote: > Dear All, > > As anticipated, here is the technical comment on our Science paper by Pat > Michaels et al. It is interesting to note that one of the "et als" is Fred > Singer. > > Michaels, Knappenberger and Singer ("MKS"; I knew the metric system had its > drawbacks!) claim that our finding of a non-significant difference between > modeled (GSOP) and observed (MSUd 2LT) lower-tropospheric trends is suspect. > They base this conclusion on two main lines of reasoning. > > 1) They contend that in the real world, ENSO has induced a trend of +0.04 > degrees C/decade in tropospheric temperatures over 1997-98. They cite > Christy and McNider (1994) and unspecified "updates" as the basis for this > number. They argue that in the GSOP experiment, the effect of simulated > ENSO variability on lower tropospheric temperatures is "near zero, > characteristic of random fluctuations". They reason that the "observed" > ENSO-induced trend on 2LT temperatures should be added to the GSOP 2LT > trend, and that this would degrade the agreement that we found between > modeled and observed lower tropospheric temperature trends. > > 2) MKS contend that the neglect of El Chichon's effects in GSOP would likewise > increase the differences between modeled and observed 2LT temperature trends. > They argue that "El Chichon had about two-thirds of the cooling effect of Mt. > Pinatubo" (I presume they refer to an effect on lower-tropospheric > temperatures here). Since El Chichon erupted in April 1982, near the > beginning of the 1979-98 period that we consider, MKS reason that properly > accounting for Chichon's effects would have yielded cooler initial values > in the lower troposphere, and hence would have increased the GSOP 2LT trend > by 0.07 degrees C/decade. > > I believe that MKS are wrong on both points, but we need to do some > work in order to address the issues that they raise. > > ISSUE 1: "THE ENSO EFFECT". > > 1a. Does ENSO induce a positive trend of 0.04 degrees C/decade on 2LT > temperatures over 1979-98, as MKS allege? This issue is not as straightforward > as it might appear. There are various way of subtracting an ENSO influence > from 2LT temperatures. Tom Wigley has looked at this in some detail. We > should check the value of 0.04 degrees C/decade that MKS cite. > > 1b. What is the effect of ENSO on the simulated 2LT temperature trends in > GSOP over 1979-97? I doubt whether this effect is zero, as MKS claim. In > fact, examination of Figure 2 (bottom panel) in Bengtsson et al. suggests > that the simulated ENSO effect on 2LT temperatures may actually be larger > than the observed ENSO effect. It is important for us to check this, factoring > an ENSO influence out of GSOP in the same way that Tom has done this in the > observations. [NOTE TO CHARLES: We need to calculate the normalized and > non-normalized SOI indices from GSOP as soon as possible!] One technical > issue here is whether we estimate the model SOI/2LT lag from the limited > data in GSOP or from the much longer control run. > > 1c. We need to refer MKS to our Figure 4, and note that in a single > 20-year period of the control run, the effect of effect of simulated ENSO > variability on lower tropospheric temperatures need not be "near zero, > characteristic of random fluctuations". > > ISSUE 2: "THE EL CHICHON EFFECT". > > Amazingly, MKS completely ignore the observed sychronicity between > the El Chichon eruption and the 82/83 El Nino. The large "net" tropospheric > cooling effect that they have naively assumed just didn't happen in the > real world. The tropospheric cooling induced by Chichon was largely > (perhaps completely) offset by the 82/83 El Nino. MKS are trying to have > their cake and eat it too. They maintain that the 97/98 El Nino had a > large positive effect on observed 2LT trends, but neglect the effect > of the 82/83 El Nino in estimating the "trend impact" of El Chichon. > > So how do we address the El Chichon issue? I think we need to: > > 2a. Attempt to quantify the effect of El Chichon on the observed trend > in MSUd 2LT over 1979-98. > > 2b. Show the "raw" MSU channel 2 and 2LT records, which do not indicate > a large post-Chichon effect on mid- and lower-tropospheric temperatures. > > 2c. Cite the recent JGR paper by Andronova et al., which estimates the > global-mean radiative forcing induced by El Chichon to be half that > of Pinatubo, and not two-thirds of Pinatubo, as MKS claim. > > 2d. Perhaps analyze experiments that incorporate the effects of El > Chichon. Jim Hansen has kindly offered to let us look at the experiments > his group performed in their 1997 JGR paper. Recall that these were > performed with four different model configurations - fixed SST, Q-flux > ocean, and with two different OGCMs. In each configuration, a number of > different forced experiments were performed. Stratospheric aerosols, > stratospheric aerosols + solar irradiance changes, etc. For each model > configuration and perturbation experiment, a five-member ensemble exists. > > The time period covered by Jim's experiments is 1979-96. Comparing results > from the fixed SST and fully-coupled experiments would be useful, but > doesn't help us too much in terms of explicitly quantifying the effect of > El Chichon on lower tropospheric temperature trends. For that, Jim would > have had to perform experiments with and without El Chichon. [What we can > say is that in Jim's fixed SST experiments with Chichon, Pinatubo and > other stratospheric aerosols, the mid-troposphere WARMS by +0.18 degrees > C/decade (Hansen et al., 1997, JGR 102, page 25,705, plate 12a, column 1). > In the Q-flux ocean experiments with stratospheric aerosol forcing, > the mid-troposphere COOLS by -0.25 degrees C/decade over 1979-96 (plate 12a, > column 3). Some part of this difference in response must be due to the > "offsetting" of Chichon's tropospheric effects by the tropospheric effects > of the 82/83 El Nino.] > > We need to keep our response fairly simple and short. So I don't think that it > will be feasible for us to look at Jim's experiments for the specific > purpose of our reply to MKS. > > I hope that our reply will involve all of us. It would be much more > effective that way. MKS will do their utmost to publicize and politicize > their criticism of our Science paper (see their endnote 8). It is a real shame > that their "analysis" of our paper trivializes a complex scientific issue. > I would very much appreciate any help or advice that you can give me regarding > our response to MKS. > > With best regards, > > Ben > -- > ---------------------------------------------------------------------------- > Benjamin D. Santer > Program for Climate Model Diagnosis and Intercomparison > Lawrence Livermore National Laboratory > P.O. Box 808, Mail Stop L-264 > Livermore, CA 94550, U.S.A. > Tel: (925) 422-7638 > FAX: (925) 422-7675 > email: santer1@llnl.gov > ---------------------------------------------------------------------------- > > ------------------------------------------------------------ > > Subject: Technical Comment on RA1046025 > Date: Mon, 06 Mar 2000 19:58:02 -0500 > From: "Stewart Wills" > To: > > Dear Dr. Santer: > > We have received a technical comment on the recent *Science* paper by Santer et al., and are considering sending the comment to reviewers to determine its validity. If possible, we would like a formal response from you that would assist the reviewers in their evaluation of the comment's suitability for publication. I have attached a copy of the comment, in PDF (Acrobat) format; please let me know if you cannot accommodate this format. > > If the comment and reply are accepted for publication, you will be given an opportunity to approve and make any necessary revisions in your reply. If the comment is rejected, a copy of your reply will be sent to the comment author. > > We would appreciate hearing from you within three weeks. > > Sincerely yours, > > Stewart Wills > > Stewart Wills > Associate Editor > Science Online > American Association for the > Advancement of Science > 1200 New York Avenue, NW > Washington, D.C. 20005 USA > Voice: 202-326-6521 > Fax: 202-408-1256 > E-mail: swills@aaas.org > > ------------------------------------------------------------ > Name: 1050048michaels.pdf > 1050048michaels.pdf Type: Portable Document Format (application/pdf) > Encoding: base64 > Download Status: Not downloaded with message Attachment Converted: "c:\eudora\attach\dian.gaffen.vcf"