date: Wed, 21 Aug 1996 17:03:59 -0600 (MDT) from: Tom Wigley subject: Re: downscaling to: Keith Briffa Dear Keith, Sorry to hear that you feel down. I thought your work was going well---too much of it perhaps? This may not help, but poor Ben is in a much worse state (but I can't tell you the details via e-mail). Thanks for the Jenkinson background, and for reminding me about what Tom Holt is doing. You may not have seen my latest message on the coefficients. Just an algebraic error on my part. I've generalized the algorithms and coded it all up for our downscaling project. For observed data, we use the 2 x 2 degree NCEP re-analyses (1970-95); but we are re-gridding these data to the HadCM 2.5 x 3.75 degree grid. This is still much finer than the direct Jenkinson case. Furthermore, we will generate the data over a large box, giving us scope to spatially filter and look at resolution effects (at some distant time in the future). It has been great having Rob Wilby here. We have opened doors to a wide range of future applications of NCEP and HadCM data: all one needs is time--or a number of clones. Cheers, Tom On Wed, 21 Aug 1996, Keith Briffa wrote: > Tom, > I got a copy of your fax to Declan. He may have replied by now but for > what it's worth I thought I'd mention that Jenkinson never intended his > formula to be used for anything other that defining Lamb types and the > Gale index. The latter was I believe 'adjusted' on empirical grounds as he > recognised the inherent smoothing problem of using a course resolution > grid. As for the Lamb types I'm sure you're right that he designed the method > specifically for the application and did want a grid area that gave results > more consistant with the types as designated by Lamb. For application anywhere > else one would have to use a degree of smoothing appropriate to the scale > of interest - or one might be constrained by the maximum resolution of the > oberved field. As for precipitation, I agree with you that ( at least for > certain types of rain) you need finer spatial resolution and your method > seems fine. This can of course be tested with real station precipitation data > and high resolution pressure data gridded at different spatial resolution. > I think Tom Holt has access to high-resolution pressure data from some Norwegian > source, or perhaps station pressure data or ECMWF data could be used. > I can't help with the coefficient problem. I remember Phil trying to > justify these numbers when we did the original work - at one time we thought > there might have been a typo in the original paper - but I remember Phil > being happy that he could justify the numbers eventually. I think he may of > contacted Jenkinson . He will , no doubt , fill you in on this - but we know > that the method works well in the context of defining the Lamb Types. This , > as you say, is not the problem. You need to do the tests with different > grids and real data to decide the most appropriate scale for different > variables. > Hope you are well . I've just come back from two weeks in Aldeburgh and I,m > feeling pretty depressed about life in general and work in particular - I have > two short of 100 email messages ! My best to Astrid and the children. > Keith > -- > Dr. Keith Briffa, Climatic Research Unit, University of East Anglia, > Norwich, NR4 7TJ, United Kingdom > Phone: +44-1603-592090 Fax: +44-1603-507784 > > ********************************************************** *Tom M.L. Wigley * *Senior Scientist * *National Center for Tel: 303-497-2690 * *Atmospheric Research Fax: 303-497-2699 * *P.O. Box 3000 E-mail: wigley@ucar.edu * *Boulder, CO 80307-3000 * *USA * **********************************************************