cc: t.osborn@uea date: Tue Jan 6 18:50:09 1998 from: Tim Osborn subject: Re: Previous message sent to Tim ... to: David Stephenson At 11:35 AM 12/23/97 +0100, you wrote: >1. I would be very interested to see your analysis >of the HadCM2 ... it might give me some ideas of what >to look at in the 18 CMIP models. If you have any >fascinating points then let me know them and we can >look for those in the other models. Things have been progressing rather slowly due to the NERC proposal and other commitments (hence my very delayed reply), so I have yet to complete a draft write up of our findings. But I hope to do so very soon and will certainly send you a copy as soon as possible. I very brief summary is that: HADCM2 has an NAO, with a multivariate spatial structure very similar to that observed (in pressure, temperature, precipitation and synoptic variability), with the possible exception of enhanced correlation with what's going on over the North Pacific. Temporally it is also very similar if the recent observed trend from the 1960s to the 1990s is ignored. If the latter is included, however, then it is outside the range found in the 1400 yr of model data. It is also not explained by the climate change signals found in ensembles of CO2 and CO2&aerosol forced integrations of HADCM2, which show a decrease in the NAO index going into next century. I've yet to check individual ensemble members in detail, though, but have noticed that some appear to show a peak in the NAO index in the 1990s. >2. We have all the CMIP data now and will start analysing >it in spring 1998 ... the idea is to keep the analysis as >simple as possible. Plots of variance, filtered variance etc. >We have developed a simple box index (NAOSTI -- NAO ST index) >which seems to agree well with Hurrell's NAO index and with >PC1 of the Jones ST covariance. We will calculate that for all >models together with PC1 for each model. Good luck! Curt Covey indicated to me that most models submitting perturbed 80 yr integrations for CMIP2 would also submit SLP and precipitation control runs to go with them. That may help too! >3. It may be nice if you joined us on the CMIP comparison. Your >experience with the UKMO results would be invaluable. It would >be great to have you on board (and it doesnt involve any bureaucracy >since we dont get any money for doing NAOMIP !). Yes, I would be keen to be involved in this way. Even more so if you plan to extend the project to include SLP from CMIP2. Its probably best if you look through the draft paper that I will soon be sending you, and that may (or may not!) suggest aspects to which I could usefully contribute. >4. I am working more with real statisticians these days and would >be very interested in knowing about your NERC project and being >included if possible ?? I could help supply some statistical >methods to better extract the predictable part of the NAO signal. >Please consider me in the proposal if you haven't already sent it >off. Unfortunately there was no time for this - sorry. The two PIs (Keith Briffa and Phil Jones) managed to be out of the country for the last weeks of November immediately prior to the 1 Dec submission deadline. It meant that it was all a bit rushed to get the proposal finalised. >5. I work with some Spanish scientists who are very interested in >relating their rainfall to the NAO. Yesterday, they explained to me >that they have a lot of station data (>100years monthly means) over >the Iberian peninsula except over western Spain and Portugal. Do you >have rainfall data sets for these regions ?? Personally, I do not. Mike Hulme's gridded precipitation dataset ( see http://www.cru.uea.ac.uk/~mikeh/datasets/global/ ) has monthly gridded values and he may be able to supply you with the underlying monthly stations values. For use only by the MEDALUS project ( see http://www.cru.uea.ac.uk/~jeanp/med1.htm ), Jean Palutikof and Clare Goodess have access to some daily and monthly station data. While they cannot give you access to these data, they may be able to point you to wherever they obtained them from. >P.S. As part of CMIP1, there would be no problem getting hold of >SLP data from some of the larger groups such as UKMO, GFLD, NCAR, MPI. >(GFLD=GFDL). See above re. CMIP2. >P.P.S. What do you know about NAO 2-5 year variability and the >possible quasi-bienniality of the NAO ?? Any ideas welcome ! Not alot. I do know that Ed Cook has done some Singular Spectral Analysis of the NAO index and identified quasi-bienniality. It also appears that Ed's reconstruction of the winter NAO index back to 1700 using tree-ring-width data captured variability on these timescales better than on some other timescales. Regards Tim