cc: Phil Jones , Kate Willett , Dick Dee , david.levinson@noaa.gov date: Fri, 30 Jan 2009 10:41:06 +0000 from: "peter.thorne" subject: Re: CRUTEM3v and HadCRUH comparisons with ERA-40 and ERA-Interim to: Adrain Simmons Hi Adrian et al., thanks for the update. Looks good. Very interesting stuff. I'm including Dave Levinson in here to speak to the precip issue in my very last point. I knew there had to be some benefit to my being chapter lead in the BAMS State of the Climate report. Dave did the precip analysis for AR4 or at least can get hold of the different datasets. Okay, on with comments: trying to head off likely reviewer concerns in tables 1 and 2 is quoting to 4 s.f. scientifically justified given the likely errors that are involved in both the obs datasets and the reanalyses? Would 2 or at most 3 s.f be more rational? Is a difference in the 3rd or 4th s.f. meaningful and pointing to a real difference that we should be investigating or simply an artifact of inclusion / exclusion of one or two datapoints? I like the plotting style in Fig 1., can this be replicated in Figure 6 so we are internally consistent? I'd still like to see an assessment of the significance of differences when we compare obs and rean timeseries (various places) in a formal sense. Maybe this is in the text but not the Figures in which case please move on move on, nothing to see officer. Otherwise I just feel that this protects us from an obvious angle of attack by the usual suspects if this thing gets picked up and turned over by climatefraudit or associates. Saying its statistically indistinguishable also bolsters the scientific value of the analysis substantially in my view. I'm intrigued by what is going on in sub-saharan Africa ERA-INT in Fig. 3. Do we have any confidence in this feature? It kind of jumps out at you so may be worth addressing / discussing. Figure 11 and associated discussion seems a reasonable strawman to me. What does the rean soil moisture do as that is the local source over land. Do we see large-scale drying? Pity we don't have a reasonable soil moisture obs dataset. I think in the rean at least we have access to enough variables to actually corroborate the strawman even if we cannot then confirm it in the real-world because of a lack of data ...! In figure 12 you use GPCP data. Now precip estimates over land are highly uncertain, and differ substantially on multi-decadal timescales. So I'd be loathed to rely upon a single dataset version to make a meaningful conclusion about the differences between obs and rean. 1999 would fit rather well with ATOVS introduction and that may help explain any apparent jump. But equally plausibly GPCP may be in error. I'd at least use the NCDC GHCN dataset and possibly others here to be totally sure that we are attributing the problem to the right cause (rean or obs). Thanks Peter -- Peter Thorne Climate Research Scientist Met Office Hadley Centre, FitzRoy Road, Exeter, EX1 3PB tel. +44 1392 886552 fax +44 1392 885681 www.metoffice.gov.uk/hadobs