date: Tue, 06 Jul 1999 15:33:18 +0100 from: Frank Oldfield subject: Impact Assessment. to: k.briffa@uea.ac.uk Dear keith, Hope all is going well for you. Rick suggested I pass this harangue of mine on to you. I've shrtoened it a bit (believe it or not). It reflects more frustration than inspiration, but if it rings any bells with you, maybe it is worth not forgetting: "The more I listen to the current crop of Integrated Impact Assessment (IIA) and related stuff (I volunteered for another stiff dose at the Boston AGU and held on up to the point when aversive therapy kicked in), the more impatient I get. With very few if any exceptions, they fail to make adequate or intelligent use of the paleorecord even if they show any awareness of it at all. If we really believe in what we increasingly see as shifts in modes of variability/ extremes beyond the instrumental record/human activity-climate variability interactions/ non-linear and relatively irreversible responses to threshholds etc etc all from evidence that lies beyond the range of instrumental records we should be pushing harder to make the IIA community attracted to and interested in making use of what we have to say. It seems to me that one way to break in would be through the current concerns for sustainability and food security etc. What seems mostly to happen at present is the development of scenarios from coupling three kinds of model: (i) GCM-based climate projections reflecting 2XCO2 or whatever, downscaled in some way to regional level, or alternatively, synthetic climate variability produced by a weather generator type of model. (ii) crop models based on physiology etc. (iii) socio-economic models which themselves aim to couple macro (national/international) influences and small scale (eg.individual farmer) decision making processes. There's not much we can do about (ii) or (iii), but we can at least ask ourselves whether there is any alternative to (i). As I see it, future climate scenarios are pretty ropey at global scale anyhow and they do precipitation and moisture balance (key properties for crops etc) badly relative to the level of confidence needed. Once they get downscaled to regional level, they are worse. The alternative approach, to use 'weather generator' type modelling capability, may be better in some ways, but relies entirely on relatively short instrumental records at best as far as I can discover. In my view, the key 'strengths' of this kind of approach are (i) the relative fluency with which scenarios can be developed (ii) total lack of any way in which they can be tested except in real time (by which time it is too late and one can always claim a new and better model is now available that accommodates the problems that screwed up the last one) (iii) existence of a powerful priesthood with a vested interest in the whole business. NOTE - these are exactly the very characteristics that kept the Oracle at Delphi in business and my best guess is that those Greek ladies probably did better though the odd war went astray! Contrast the above with paleodata which gets better at the region specific level and actually measures reality. If we look at Golbal or pan-Hemispheric reconstructions of climate variability over the last few centuries, the amplitude of decadal or century scale variation falls short of that projected for the next century, so perhaps it is difficult to push the idea that on that spatial scale future scenarios can be effectively fuelled by paleodata. But if we look at regional level, the variations are aften larger and for extreme years/seasons/events - whether they actually occur serially or not - it is often possible to develop credible synoptic situations linked to the phenomena. If then we think about future scenarios not as changes in hypothetical means but as changes in the frequency and persistence of documented 'anomalies' we can generated scenarios that are both based on reality and make use of the full range of variability we can reconstruct. Of course this approach too has its flaws and limitations, since future forcing may generate weather patterns etc. outside the range of past extremes and so forth, but I do not think these flaws are anything like so severe as the ones that afflict the other approaches. The big question is ' is there anything we can do about this?' . What I feel we really need are contacts at the potential 'user' end who are neither locked into promoting climate scenario/model development for their own advantage nor closed minded to the possibility that other appraoches could be useful. Any ideas? Frank ____________________________________________ Frank Oldfield Executive Director PAGES IPO Barenplatz 2 CH-3011 Bern, Switzerland e-mail: frank.oldfield@pages.unibe.ch Phone: +41 31 312 3133; Fax: +41 31 312 3168 http://www.pages.unibe.ch/pages.html