From: Phil Jones To: Tom Wigley Subject: Re: T data Date: Mon, 28 Oct 2002 15:50:07 +0000 Cc: Ben Santer ,t.osborn@uea.ac.uk Tom, Talked to Tim re the SD field. Can you read the following (J. Climate 10, 2548-2568) before you come so you know how Tim infilled the SD field ? HadCM2 data was used. This would seem to bias any model validation to this model. Also it would seem odd to validate any model in a region where there is no data - in a region that had to be infilled. I can see that global fields make things simpler, but they will need to constructed in the best possible way. In 1997 we thought the best way was to use a model, but our aim then was different from yours. Cheers Phil At 06:04 28/10/02 -0700, Tom Wigley wrote: >Phil, > >Thanx. I need to see if CMIP has the height fields for models --- >Ben???? > >Tom. > >_______________________________ > >Phil Jones wrote: > > > > Tom, > > Here's the file that you should have got back in September. It is > > 1981-2000 where this > > could be calculated and 1961-90 elsewhere. The other fields (already > > sent) enable you to > > know where the 1961-90 field has been used. > > All you need to overcome the problem of this being surface > > temperatures is to get a > > 5 by 5 degree average height field. I have emailed Mark New to see if he > > has a 1 by 1 degree > > height field, which could then be averaged. Mark must have had this at > > some stage - he > > has a 10 minute height field for the world, which I'm sure he has > > degraded to 1 degree. I > > have a land/sea mask at 1 by 1 degree, so am hoping Mark has the heights. > > With this > > all you will need is the model height fields. > > As for the SD's it would be possible to produce this for a period > > like 1981-2000 or 1961-90 > > but both would have gaps - probably exactly the same as in the > > climatology. The options > > to consider here are: > > > > 1. Period 1981-2000 or 1961-90? > > 2. How many years in each needed to get an SD? > > 3. How to infill the gaps? > > > > Tim Osborn must have infilled the gaps for the errors paper in 1997 as we > > needed a complete > > field of variances. He did this by blending some model data > > (HadCM2/ECHAM3 probably) > > with the real observations. Most areas get infilled easily - big problem > > is the Southern Oceans > > and the Antarctic (also central Arctic). I will talk to Tim. > > > > We can discuss this more when you come. > > > > Cheers > > Phil > > > > PS I should have some results from Anders by the time you come. He is > > comparing means/ > > SDs and extremes etc of HadRM3 with real world data from 200 sites across > > Europe. Only > > temperature variables in the first part. Clearly shows that for > > islands/coasts comparisons > > must be with land points in the model. We've had to 'move' some stations > > to be on model > > land to get better comparisons. Islands that are not in the model have > > poor comparisons. > > It is possible to see country outlines in some comparisons with either > > max or min > > temperatures. Corrections for elevation are needed to get over large > > elevational differences > > between stations and the model, but the Alps are still visible. Lapse > > rates work best only > > in some seasons - not very good in summer. Max temps produce consistent > > difference maps > > (model-obs) over Europe, but mins are more erratic/random. Min error is > > overall small but > > with a large variability while max has a larger error but low > > variability. Due to mins being more > > affected by local environment. > > > > At 09:13 27/10/02 -0700, Tom Wigley wrote: > > >Phil, > > > > > >Re my last email .... > > > > > >I have looked at the data you sent. It would be very nice to have a > > >gapless 1981-2000 T climatology to match the Xie/Arkin precip > > >climatology. However, this means somehow filling in the gaps in the > > >61-90 minus 81-00 differences, a nontrivial task. So my choice in the > > >absence of this is either a gappy 81-00, or a full 61-90. I have chosen > > >the latter -- perhaps we can discuss how to produce a gapless 81-00 > > >climatology when I am at CRU? > > > > > >A problem with the 61-90 is that it is surface, and that observed > > >surface is not equal to model surface. I'm sure you have thought about > > >this (in the model validation context) already, so this is another item > > >to discuss. > > > > > >For precip, I also have the inter-annual S.D. climatology, so I can > > >validate both the mean climate and the variability. Very interesting. It > > >would be nice to be able to do this with temperature (especially since > > >the mean climate for temperature in the models is pretty darn good -- > > >but how good is variability?) Is there an S.D. climatology for > > >temperature that you can send me? > > > > > >Cheers, Tom. > > > > 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 > > > ---------------------------------------------------------------------------- > > > > ------------------------------------------------------------------------ > > Name: newabsref8100.out > > newabsref8100.out Type: Plain Text (text/plain) > > Encoding: base64 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 ----------------------------------------------------------------------------