cc: david.parker@metoffice.gov.uk, Jim Renwick date: Thu Dec 2 09:10:44 2004 from: Phil Jones subject: Re: Layout/fonts etc for Chapter 3 to: Kevin Trenberth Dear Jim and Kevin, Have had a quick look at sections 3.4 and 3.6 that will help me with 3.9 this weekend. I have two comments on aspects of 3.6 1. In the box that defines the indices, each should give the first year of the series we'll be using (assuming all get updated in nearish real time). Also there should be one or two alternates for each series referenced. For the SOI, all basically use Tahiti and Darwin in one form or another. Refer to Können, G.P., Jones, P.D., Kaltofen, M.H. and Allan, R.J., 1998: Pre-1866 extensions of the Southern Oscillation Index using early Indonesian and Tahitian meteorological readings. J. Climate 11, 2325-2339. as this gives the SOI back to 1866 (and earlier). For the NAO, the Hurrell (1995) index is based on Lisbon and Stykkisholmur, Iceland (as sort of said in the caption to the Figure of SLP indices for Dec-Mar). The caption says Stykkisholmur/Reykjavik. It should be one or the other ! I can get both to Jim Hurrell or whoever's doing the diagram. I prefer Reykjavik as it easily updates itself in real time. Reykjavik is available (daily, by the way) from 1821. In the text it says Akureyri but that doesn't start until about 1880. In the text refer to Jones, P.D., Jónsson, T. and Wheeler, D., 1997: Extension to the North Atlantic Oscillation using early instrumental pressure observations from Gibraltar and SW Iceland. International Journal of Climatology 17, 1433-1450. This is in the reference list. This is based on Gibraltar and Reykjavik and goes back to 1821. Although Hurrell's 1995 index is probably the most widest one used, the one with Gibraltar is catching up. I've reviewed a couple of papers over the last year (which won't be seeing the light of day) which said they used Hurrell back to 1821. They were using the one based on Gibraltar without saying. The rest of the papers were poor as well ! One possibility with all the indices is a small table of correlation matrices between those we discuss/use. All to show that the exact choice of index doesn't make that much difference. For the long NAO series it is important to highlight how unusual the 1965-95 period is wrt the whole period from 1821 (see also point 2 later). The traditional NAO index is based on Ponta Delgada and Stykkisholmur. This goes back to 1865 - referred to in Jones et al (1997) and in Jones et al (2003) in Hurrell's book. This index is also the one essentially used since Walker's work (i.e from the 1930s to 1995). I think Jim Hurrell would have used this if he'd realised that the Azores existed back to 1865 and not 1895 as it did in NCAR and GHCN archives. The Azores site is better for the rest of the year (non-extended winter season) than an Iberian station. I'm glad that Jim H did this as I wouldn't have worked on the Gibraltar series. The paper in 1997 has over 200 citations and only the global temp papers get more for me. We should have this back daily to 1821 as well soon, by the way. Also, through an EU project (EMULATE) we will be able to extend the PC version of the NAO back to 1851 as we have a daily MSLP dataset for the N. Atlantic and Europe back to 1851. Paper on this should be submitted early next year. 2. In the NAO section (3.6.4) I agree more should be said about what the NAO does for surf T/P and storm tracks etc (and similarly for the other indices), but much more should be made of the variability of the influence of the NAO over time ( a la Jones et al 2003). This variability of influence should be discussed for the SOI as well. It likely occurs for the other indices as well, but records are shorter. For the SOI it has been documented by Rob Allan and others. When pointing out that the NAM or the NAO accounts for some of the warming of Eurasian winter temps, Thompson (2000) is misleading as it just uses the best period of 1965-95. If these sorts of calculations of variance explained (and trends ) get repeated over much longer periods the NAO explains variance of UK, Fennoscandian temps over long periods, but doesn't explain much of the trends. It only appears to over 65-95 because the NAO has a trend. For the NAO and the Eurasian temps, this reduction in influence for certain earlier periods has nothing to do with data quality - all are OK back to at least 1851. Similarly with the SOI, this explains variance of large-scale temperatures (NH and SH averages) but it doesn't explain any long-term (100 yr) trend. It makes the residual trends more significant, but this is because some high-freq variance has been explained. The SOI has no long-term trend. Prior to the 1930s some of the variability of the influence may be due to slightly poorer data, but this is a second-order effect. The NAO barely has a long-term trend over 1821-2004, so in turn it can't explain long-term trends at the century scale. So I'd like to see expansion of the caution expressed. A number of the skeptics say the Eurasian winter warming is due to the NAO. Only a small part of this is true. It is important to get this across as the last 5-6 winters have had near zero NAOs, but all have been mild on average in the UK and Fennoscandia. I have a few ideas why the influences change, but have never put them into print. One is anthropogenic influences, but this is difficult to prove. The point to get across is that they do. Cheers Phil Prof. Phil Jones Climatic Research Unit Telephone +44 (0) 1603 592090 School of Environmental Sciences Fax +44 (0) 1603 507784 University of East Anglia Norwich Email p.jones@uea.ac.uk NR4 7TJ UK ----------------------------------------------------------------------------