cc: Gabi Hegerl , dkaroly@ou.edu, myles , Tim Barnett , Nathan Gillett , Phil Jones , Jesse Kenyon , Reto Knutti , Toru Nozawa , Doug Nychka , Claudia Tebaldi , Ben Santer , Richard Smith , Daithi Stone , "Stott, Peter" , Michael Wehner , Xuebin Zhang , francis , Hans von Storch , Thomas R Karl , "Bamzai, Anjuli" , Chris Miller date: Wed, 02 May 2007 13:57:02 +0100 from: peter.stott@metoffice.gov.uk subject: Re: 3 things, please reply by May 2 to: Tom Knutson Hi Gabi and everybody, I agree with all the main points made so far. I think it would be a good idea to future-proof our "past" simulations (as we did with our current HadGEM1 simulations calling the "past" segment up to 2010). The forcings for this predictive part of the "past" simulations could follow the business as usual path advocated by Nathan and for solar we can carry on the solar cycle and for volcanoes assume no big ones (if we get one we'll have to re-run of course; or if we have one between now and when we start these expts decay the aerosol). So for AR5 having attribution ensembles available that go up until at least 2010 would be a good idea I think. We would like also to have initial condition ensembles for d+a of course. Ideally also we would have separate forcings simulations to at the least separate out the ghg component from the other anthropogenic factors (to separate out TCR from forcing uncertainty) and separate natural forcings simulations. Any dates for the meeting are fine with me - by that stage IPCC SYR will have finished - wey hey. Can we go skiing ? Best wishes, Peter On Tue, 2007-05-01 at 16:49 -0400, Tom Knutson wrote: > Gabi et al. > > I don't have any further recommendations on overall design. I agree > with David and Nathan that the transition between past and future > segments of the runs needs to be handled more cleanly than was done > in AR4, allowing analyses to be extended to some extent as more years > of observations become available (i.e. as time marches on...) > > I also agree that a fairly large number of ensemble members will be > needed, particularly to deal with the issue of large multi-decadal > internal variability, such as the AMO. > > Regarding the additional upper-air variables to save, here are the > variables that we are currently using to identify tropical storms in > our regional model at GFDL. > Although there are a number of methods of detecting storms in use, > that may use slight variants on these, this at least gives a flavor > for what is typically used: > > 6-hour time resolution grids of: > 1) 850mb U, V > 2) lowest model level or 10-meter U,V > 3) Sea level pressure > 4) Temperature at 300mb, 500mb, and a few levels in-between > (in order to compute vertically averaged temperature in > the 300-500mb layer). > > Concerning meeting times for next winter/spring I have no > preference among the choices put forward. > > -- Tom Knutson > > > David Karoly wrote: > > Hi Gabi, > > > > 2) I think that the new coordinated model experiments should have high > > resolution runs starting in 1970 (giving more than 30 years of obs data > > for d&a analysis) and running until 2040. We want to also stress that we > > would like different forcing combinations at least for the > > 1970-2005/2010 period, to allow attribution to different forcings. > > The high resolution runs need to NOT be cold start but to be spun up > > either from observed ocean data or from lower resolution coupled ocean > > atmosphere runs with data assimilation. > > Also, for lower resolution, we need to have some runs starting in 1880 > > or 1900. > > > > If we don't have long control runs at high resolution to estimate > > climate variability, we may need more ensemble members, or clever ways > > to estimate decadal climate variability. > > > > We should also make some IDAG recommendations on the specific variables > > that should be saved that are in addition or different from the current > > ones on the AMIP CMIP3 archive, like monthly mean Tmin and Tmax, or > > daily near surface specific humidity, as well as high time and space > > resolution surface and upper air data etc. > > > > Gabi, I can't go to the WGCM meeting in Hamburg in 3-5 September, as I > > have other meetings in Oz and the US around this time. Will you be able > > to go, as it will be important that the d&a community is represented there. > > > > 3) In terms of dates, I will likely be in the US for the AMS annual mtg > > during 20-24 January in New Orleans. If I stay in the US, I would prefer > > the IDAG mtg to be as early in Feb as possible, around 9 Feb, or as late > > in March as possible if I go back to Oz in between. > > > > Best wishes, David > > > > Gabi Hegerl wrote: > >> Hi IDAG people, > >> > >> Three things: > >> > >> Jesse has put the collection of powerpoints from our meeting > >> (those that people were not uncomfortable to > >> share) on a webpage, instructions below, you are welcome to find talks > >> and get them. > >> > >> Secondly, I attach Jerry Meehls writeup on the planned AR5 > >> experiments, it would be > >> helpful if we could comment on them as group. Please send comments to > >> me, and I will > >> collect and circulate our group reaction before sending to Jerry. My > >> personal view is > >> that the high-res near future runs are a great idea, but should start > >> early enough for > >> us to do some high res attribution, so eg 30-50 years over the 20th > >> before going into > >> 21rst. > >> > >> Also, Doug says that if we would soon decide on our next meeting > >> timing, we may be able > >> to get a particularly attractive location at NCAR (forgot what its > >> called). We tentatively > >> planned some 3 day window between February 8 to march 16. Myles points > >> out that > >> Brits wanting to bring kids would do well with dates on either side of > >> the weekend 16-17 February for the Southern part, and 9-18 for the > >> Northern part (which is when school > >> is out). > >> Greetings > >> > >> Gabi > >> > > >