From: "Griggs, Dave" To: 'TAR CLA list' , 'TAR LA list' Subject: Uncertainties again Date: Fri, 08 Sep 2000 18:02:09 +0100 Cc: 'TAR Review Editors' , "'Watson, Bob'" , "'Moss, Richard'" , "'Houghton, Sir John'" , "'Albritton, Dan'" , "'Swart, Rob'" , "'Leary, Neil'" , "'McCarthy, Jim'" , "'Stone, John'" , "'shs@leland.stanford.edu'" , "'m.manning@niwa.cri.nz'" Dear CLAs/LAs As you all know, in my Victoria follow-up e-mail of 2 August I presented a summary of the agreement we reached in Victoria on a common use of terminology to express degree of likelihood in the TAR. At that time the word or term to be used for the central box of 33 to 66% had not been agreed and the word "inconclusive" was proposed for that category. Since that time there has been a lengthy discussion, including Working Groups II and III, regarding the best word to be used in this category. To cut a long story short the term we would now like you to use for this middle category is "medium likelihood". I am sorry I have not been able to canvas this around all of you but from the discussions this term was agreed by all to be the best compromise. In particular, it clearly maintains the scale as one of degrees of likelihood, whereas inconclusive could be confused as to whether a degree of likelihood was being expressed or whether there was insufficient information to conclude a likelihood. I attach a table showing what should now be the final scale. During the discussions it became clear that in addition to making likelihood statements it is sometimes more appropriate to express statements in terms of a degree of confidence, and indeed several chapters use this terminology. As you know the Uncertainties Guidance paper by Richard Moss and Steve Schneider recommends a scale of confidence from Very Low to Very High confidence. WGII in particular are using this scale and so I would ask that, if you choose to express things in terms of a level of confidence, that you use the terms as they are laid down in the guidance paper. This in no way affects the use of the likelihood scale where this is more appropriate. For example, if we are highly uncertain how well a model handles a particular process, we may have "very low confidence" in a model result which is highly dependent on this process. If we have no other corroborating evidence we may therefore conclude that there is insufficient information to assign a likelihood in this case. By following the guidance paper when expressing a level of confidence we will hopefully improve the consistency between the two reports. Incidentally, if there are instances in the WGII report where they are able express degrees of likelihood they are going to try and use our scale. Thirdly, there has been a lot of discussion about the impression which the likelihood scale, if taken out of context, could give for low likelihood, high consequence events, such as a disintegration of the WAIS or a shutdown of the THC in the next 100years. Please bear in mind that policymakers must balance likelihood and consequence in deciding whether or not to take action. Therefore please take extra care when considering text for these types of issues as simply expressing them as "extremely unlikely" does not give the full picture. For example, you could say an aircraft was "extremely unlikely" to crash on its next flight but if there was a 1% chance I would not fly on it. While it is a true statement the right balance is only achieved when the consequence is also brought in to put the risk in context. I apologise for this late change to our scale but I hope you all agree that it is an improvement. If anything is not clear about any of the above please do not hesitate to contact me. Best regards Dave <> ----------------------------------------- Dr David Griggs IPCC WGI Technical Support Unit Hadley Centre Met. Office London Road Bracknell Berks, RG12 2SY UK Tel +44 (0)1344 856615 Fax: +44 (0)1344 856912 Email: djgriggs@meto.gov.uk ----------------------------------------- Attachment Converted: "c:\eudora\attach\Agreed terminology2.doc"