Page 14, line 21. Change to be consistent with Page 9, lines 34-36 noted above. There is no significant change in any of these records since 1979.

Page 20, lines 26-40. Way to take something we didn't say and say we said it!!! Here's what the EOS paper (Lins and Michaels, 1994) concludes in its final paragraph:

"One possible explanation for the differences between Karl's findings of no (emphasis added here for the IPCC folks) increases in winter-season precipitation and our increasing winter streamflow is that there may have been a decrease in evaporation which would occur if there were an increase in cloudiness even if temperature and precipitation did not change...[cloudiness increases in the U.S. were described by Angell, 1990]...The streamflow and cloudiness data are also consistent with the trend toward lower daily maximum temperatures (and therefore less evaporation) that occurred across the United States from 1948-87".

Please make the text consistent with the citation. Try, "Their results were most consistent with observed reductions in daytime high temperature and increased cloudiness over the United States, as streamflow increases were observed over regions and seasons where precipitation has not increased".

Page 25, lines 10-14. Mastenbrook (1983) found larger declines over the same regions noted in Oltmans and Hoffman for the 1960s and 70s. Somehow this should be noted.

Page 27, line 17. Trenberth (1995) is submitted, not accepted.

Page 29, line 20. Hurrell (1995) also submitted, not accepted. Because this is cited immediately after Trenberth and Hurrell, I'll guess that they're related. While Kevin appears to know more than anyone about climate (just ask him), even his work may be subject to revision after review.

Page 30, line 26. The 60-year "period" referred to by Schlesinger and Ramankutty is based upon one (1) very noisy cycle length. Errors in hemispheric temperature projections by Schlesinger's GCM bear little resemblance to this pattern. Thus the text should be modified to indicate that the magnitude of the Schlesinger and Ramankutty period is insufficient to explain any statistically significant portion of current transient model errors in the Northern Hemisphere.

Page 32, lines 4-5. This is a decidedly "non-greenhouse"-like circulation change in the hemisphere that is supposed to be unfettered by sulfates. Contrast this to Page 33, lines 22-23. Together you must conclude, from these observations, that the circulation in the Northern Hemisphere is more consistent with a greenhouse alteration than is the Southern Hemisphere.

Page 39, line 23. In this case, both manuscripts are "submitted". Because this reviewer cannot comment on the statistical significance of these findings, they are best left out.

Page 47, line 18. Figure 3.22 is very interesting and deserves comment. The text should be altered to state, as I noted in an earlier comment, "Composite indicators of summer temperature indicate that a rapid rise occurred around 1920, and that made the 1920s and 1930s the warmest decades since at least around 1400. This rise was prior to the major greenhouse emissions. Since then, composite temperatures have dropped slightly on a decadal scale."

"Cartoon" illustration: Only statistically significant changes in observed data should be included.

REVIEW OF CHAPTER 6

My greatest difficulty in reviewing this chapter relates to evaluation of the sulfate/greenhouse transient model of Mitchell, which was used in Berlin as to ground statements that models and observations are now much more consistent. My examination of Mitchell's upcoming Nature manuscript leads to the conclusion that this model is still producing unrealistically large warmings precisely in the region where the strongest warming is forecast: the high latitudes of the Northern Hemisphere. To perform a proper review, one would need to examine the spatial and temporal behavior of this signal. Consequently I repeatedly requested gridded transient data for this model and was explicitly refused. The grounds for refusal were not logical.

In other words, I was purposefully prevented from performing a proper review of this Chapter.

Page 6.1. Lines 1-4. This is wrong. As noted in my review of Chapter 3, there is no staustically significant trend in the radiosonde, MSU, or ground based records since their concurrence begins in 1979.

Lines 11-13. Quantitative inconsistency. It is 50 years to the middle of the next century. At a warming rate of .2 to .35/decade, the temperature rise to then is 1.0 to 1.8°, not the "between 2 and 4° warmer" given in line 13. Probably best to eliminate the statement on line 13.

Lines 37-38. The exact same mistake is repeated. .15 to .25°/decade gives a warming of from .75 to 1.25°, not the "between 1 and 2.5°" given in line 38.

Page 6.5 lines 2-3. This sulfate aerosol forcing change implies that annual unscrubbed coal combustion will double by 2050. That is doubtful on several grounds.

Page 6.5. The text should note somewhere that the Cubash et al (MPI) transient model (given as MPI CO2 anom) in figure 6.2.2 is clearly making major (approximately 6°C) errors, by simulating the high latitudes of the Northern Hemisphere too cold for roughly

its first 65 years. If this massive error were eliminated, the warming it predicts would begin long before now and there would be a considerable disparity between this GCM and reality. If this is not noted in the text, it surely will be in the public discussion following release of this report. Why not save IPCC's credibility by doing so now?

Page 6.6:

My reading of Mitchell's upcoming Nature manuscript leads me to believe that transient greenhouse/sulfate models continue to produce large warmings of the high latitudes of the Northern Hemisphere that have not been observed. To verify this in space and time requires analysis of the gridded output from this model, which was denied to me. This has severely compromised my ability to properly review this chapter.

Page 6.9, lines 3-4.

Again, Mitchell's Nature manuscript still implies a large winter warming should have taken place that is not apparent from observations. I cannot comment further because I do not have the transient output.

Pages 6.10-11.

The magnitude of flux adjustments, as noted in Murphy and Mitchell (1995, Journal of Climate) in the western oceanic gyres should be explicitly stated here.

Section 6.6.1.

It is clear that regional temperature and precipitation biases are so large as to make such estimation of future climate a dangerous exercise. Why include this if it is so unreliable?

COMMENTS ON SUMMARY FOR POLICYMAKERS

SPM-1, line 8. Several opinion polls of scientists have concluded that the use of the world "already" conveys a (purposeful?) misleading impression to the (mostly) non scientists who will only read this chapter. Remove the world "already" and this criticism will not be leveled.

SPM-2, top section. I do not believe in the "basic integrity" of the models because of the large errors that are currently made in high latitude winter. This is where the greatest warming is forecast and where there has been little evidence that this forecast was conccl. You should add a "bullet" after line 18 that states,

°continued overestimation of high latitude winter warming is a serious problem that has generally eluded correction, even after allowing for competing effects of anthropogenerated aerosol.

SPM-3, lines 18-19. "in recent decades" is misleading. It conveys the impression that the composite temperature indicator (Figure 3.3) is at its highest value; it clearly is not, and the highest values were reached in the 1930s with the big rise between the 19-teens and the 20s; look at the figure. For accuracy, the text should read, "Composite indicators of summer temperature indicate that a rapid rise occurred around 1920, and that made the 1920s and 1930s the warmest decades since at least around 1400. This rise was prior to the major greenhouse emissions. Since then, composite temperatures have dropped slightly on a decadal scale." Again, take a look at figure 3.3 carefully.

Lines 24-26. These are wrong. SOI values 1990-1995 are not distinguishable from 19101917, a slightly longer period. The statement about the behavior between 1990-95 must therefore be deleted; leaving it in will compromise the scientific credibility of the document because critics will point to the similarity between the two periods (in spite of the fact that the global temperature was lower and that the transition from 1917 to 1920--after the prolonged El Nino ended--was perhaps the most notable warming period in the entire instrumental record).

SPM 11-Lines 5-6. In order to protect credibility, it's always advisable to tell the whole truth. Therefore the statement "This estimate is smaller than that.." should be changed to, "This estimate is 50% less than that given...."

SPM-13, insert in line 24:

"Radiosonde, Microwave Sounding Unit (satellite), and surface-measured temperatures show no statistically significant change beginning in 1979. After adjustment for volcanism and El Nino, the satellite data continue to show no significant trend."

SPM-13, lines 34-35. See comment on SPM-3, lines 18-19, above. It is very important that you state this accurately or criticism may be very sharp.

SPM-14, lines 12-13. Same as SPM-3, lines 18-19.

Lines 39-40. The repeated citation of in the entire document, including the SPM, of *10.000 years" concerns me. Clearly, the addition of another 1,000 years brings us back to periglacial conditions, and therefore larger climate oscillations generally associated with a colder climate. Why just stop at 10,000? Because the ecological effects of natural climate change 11,000 years ago dwarfed anything humans can do??

SPM.15, lines 29-30. Most scientists would associate stronger extratropical cyclones with an increased temperature gradient; the opposite is predicted for an enhanced greenhouse effect. Add the statement, "This observation is not consistent with greenhouse projections that reduce the pole-to-equator temperature gradient. Further, there is no significant change in strong extratropical cyclone activity; intense tropical cyclone activity in the Atlantic has decreased significantly".

SPM-21, 38-40. Suggest you look at Michaels et al., (1994; Technology 331A, 123133). Sulfate aerosols are not a sufficient explanation for the known pattern mismatch between nonsulfate greenhouse GCMs and observations.

SPM 22, 38-39. Add, "There are also broad areas of considerable inconsistency between sulfate-greenhouse models and observations; these areas are the same regions where these models predict the largest warmings". This, of course, refers to the polar regions, especially in winter.

SPM 23, lines 5-13. This paragraph is wrong and will harm IPCC if it remains. MSU satellite data show no signal that is at all consistent with a sulfate modification of greenhouse warming. Critics will surely point this out as evidence for sloppy science if it remains in this Summary. Best leave the whole paragraph out rather than try to explain away the MSU with a bad argument.

Lines 14-15. Asking for trouble here. By citing the "natural variability" argument heavily in previous editions, and now switching to the absolute statement that "global-mean temperature changes of the last century are unlikely...." you're liable to be severely criticized. Best to modify the statement by saying "Some of the global-mean......”

SPM-26, lines 1-2. Previous reports said CO2 and non CO2 greenhouse contributions were roughly equivalent. You need to explain the change here (which will be very difficult).

SPM-29, lines 1-2. Chapter three notes that there is no evidence for an increase in extreme high temperatures while there is a decrease in the severity of extreme lows. You need to differentiate between these events here.

SPM-30, lines 30-31. This statement has been repeated so many times, it may ultimately be known as "Trenberth's Folly". See previous commentary. Leave it in as is and there will be sharp criticism.

SPM 34-35: Regional estimates. Delete. They are unreliable. IPCC should be aware that the unreliability of these selfsame predictions from previous editions has served as the basis for overturning or refusing externality costs for CO2 emissions in civil proceedings in several states in the USA. Leaving them in therefore harms environmental protection (if one believes that emission reductions are desirable, as does the head of IPCC).