date: Fri Jul 14 15:09:53 2006 from: Tim Osborn subject: Re: [Fwd: Research Grant Reference: NE/E002412/1 - Research Grant to: N.Gillett@uea.ac.uk Hi Nathan, thanks for offering to help out. I plan to write a reponse on Monday and will then give it to you for editing. In the meantime, if you have any views on what main points we should emphasize, then please let me know. Note we've also just received comments from a 4th referee, which is also very favourable but I'm not quite sure how to respond to their only major recommendation. Cheers Tim At 21:08 12/07/2006, you wrote: Hi Tim, Looks like the review comments on the project are pretty favourable - this is good news. I'm happy to help out with the responses in any way you want. Cheers, Nathan ---------------------------- Original Message ---------------------------- Subject: Research Grant Reference: NE/E002412/1 - Research Grant Assessments for Comment From: jrog@nerc.ac.uk Date: Tue, July 11, 2006 4:42 pm To: t.osborn@uea.ac.uk Cc: n.gillett@uea.ac.uk -------------------------------------------------------------------------- Dear Dr Osborn PROJECT TITLE: Identification of changing precipitation extremes and attribution to atmopsheric, oceanic and climatic changes Your application for the above Research Grant will be considered at the Flood Risk from Extreme Events (FREE) meeting. Referee comments have now been obtained on the application, and have included the following points to which you have been invited to respond. These comments are presented to you for clarification and do not represent any pre-judgement of the outcome of your application, and may be untypical of the general tone of review received. Should you wish to respond to the points raised by the referees, please do so by 18 July 2006 to allow us to include your response in the papers sent to the panel reviewing your application. We realise that this is a short period of time and would encourage you to submit your response by e-mail. Please make sure that you include your grant reference number. Please note that NERC is now introducing a page restriction on the length of your response. Your reply should be as concise as possible and must not exceed an average of one side of A4 for each referee assessment. For example, should you receive 3 sets of referee comments, your response should not exceed 3 sides of A4, with a minimum text font size 12. Please note that any response longer than the page limit will not be put to the Panel Meeting. This e-mail has been copied to all the Investigators on the application in case the Principal Investigator (PI) is out of contact. However, please note that we can only accept one response to these comments (preferably from the PI) - not one from each Investigator contacted. Should we receive further referee comments we will endeavour to forward these to you for comment before the meeting. Please note however, that these late comments may have alternative deadlines for response, due to meeting paper reproduction schedules, which you should attempt to meet wherever possible. No guarantee can be made that late responses will be included in the main set of meeting papers. Yours sincerely Jan Rogers Research Grants Team NERC Tel: (01793) 411574 Fax: (01793) 411545 E-Mail: JROG@wpo.nerc.ac.uk Please refer to [1]http://www.nerc.ac.uk/funding/contacts.shtml if you need to contact individuals within NERC. Referee A Sound, carefully planned scientific methodology, enabling the advancement of scientific understanding in this topic area. I view this research to be a fundamental step in refining our ability to use, with confidence, regional projections of precipitation (both mean and extreme) based on R and G CMs: as stated in the research, simulated projections of precipitation are quite uncertain - therefore focused investigation into a comparison of the mechanisms of extreme real-world and simulated rainfall is an obvious prerequisite to then using model projections to inform policy/planning decisions. The proposal to compare relationships within a variety of climate models, and the investigation of SSTs as a driver, provides an option to extend this (or subsequent) research to benefit the seasonal-to-decadal forecasting community. However, I note this is note a stated primary goal of the research. Particular merit should be noted to the naming of specific end users (both personnel and institutions). This would suggest that knowledge transfer to appropriate partners has been carefully considered and seems an important component of the research plan. Moreover, the proposal seem to indicate that some correspondence has occurred to investigate the usefulness of proposed work, so that the proposal could be tailored to benefit stakeholders. Again this is worthy of merit. I am confident that the stated objectives can be met and that the named applicants are suitable to carry out the research. Questions/hypothesis are clearly defined. I am not aware of duplicate research. I do not see flaws in the proposed scientific method. Presumably, having identified which R or G CMs most resemble the real world (in terms of humidity-circulation-precipitation relationships) and identifying the causes of any changes, the actual projected precipitation changes themselves from those models most successfully validated would be explicitly flagged to the user community-either in the literature or other avenues which might be more accessible to non-technical stakeholders? (The proposal does state that validation results could be used to weight individual models in forthcoming ensemble-type probabilistic predictions, but what about the distribution/dissemination of existing data?) I do not think that any of the resources requested are insufficiently justified. Referee B I think there is very good potential for this project to enhance the body of knowledge on the causes of regional precipitation variations observed over the past several decades. The PI's are highly experienced in the analysis of precipitation and circulation variability and in climate change detection, and they have world-class reputations in these areas. They propose to work in a part of the world where adequate data are available, where models have relatively good performance in simulating precipitation and its extremes, and where circulation/precipitation relationships have been well studied. I would therefore expect an important incremental step in knowledge. I would also expect wider applications for the study of precipitation elsewhere in the world, for application in statistical downscaling (e.g., by better informing which circulation predictors should be included in downscaling), and for application in impacts research (e.g., through the improvement of weather generators that are based on downscaling results). What are the proposal's particular weaknesses? I worry a bit that the PIs may encounter some difficulties in interpretation given that multiple proximal causes of precipitation change such as circulation and humidity change may themselves be linked, and may both be linked to anthropogenic forcing. Nonetheless, it may well be the case that S/N ratios can be enhanced in this way. I think the implications for changes in flooding risk will have to be carefully considered. One could imagine the evaluation of change in flooding risk conditional upon certain circulation influences which in turn are affected by external forcing. However, I think considerable care will be required in selecting the appropriate covariates, and in developing appropriate statistical models that do not over extend the available data resources. I didn't see much in the proposal that worries about this aspect of the research, so I must assume that all of the statistical models will be properly cross-validated. Perhaps an initial step would be to conduct a perfect model study to consider what kinds of relationships might be identifiable and the extent to which statistical models describing these relationships that do not over-fit the available data resources can be built. These are first-rate researchers with excellent statistical skills who will definitely provide a useful increment on understanding of influences on precipitation extremes and flooding. There is however non-negligible risk associated with the study of external influence on precipitation extremes given the weak signal-to-noise ratio. Success in this regard hinges on whether the disaggregation strategy will be successful. I think the request for resources was well justified. I perceive it as being basically bare bones, seeking support for a post-doctoral fellow. Given the complexity and scope of the work envisioned, it would not be reasonable to try to proceed with studentships. I would also anticipate that the pdf will get excellent supervision by the PIs. Referee C The greatest strength of this proposal is its combination of sophisticated statistical techniques for the analysis of extremes with analysis of the atmospheric factors that are related to precipitation. It is not enough to know how what changes GCMs and RCMs predict for future precipitation; it is also useful to know how well they simulate variability in the factors that affect precipitation, and how well they simulation the relationships between these factors and different aspects of the distribution of precipitation. The separation of the analysis of GCM and RCM output into these two parts will allow the proposed study to gain useful information about what these models do well and what they do not, and to use what the models do well to inform projections of future variability and change in precipitation extremes. The approach is novel, as covariates are not yet widely used to estimate GEV or GPD parameters in extreme value analysis, and the PI and co-PI have the expertise in precipitation and atmospheric processes needed to find proper covariates to use for these extreme value distribution parameters. The approach to isolate the variability in precipitation due to variability in the large-scale atmospheric circulation, developed in part by the PI, is also novel and appropriate, as it is expected to allow the detection of climate change signals in the residual once the part due to circulation variability has been removed. In addition, the project is set up well to succeed. The PI and co-PI have expertise in both atmospheric processes and extreme value theory, so they are well-suited to lead a project that integrates knowledge and techniques from both disciplines, which is required for such a project to succeed. The proposal is well thought out, addressing important issues in the application of extreme value theory to the relationships between precipitation and atmospheric properties, considering methods of testing the ability of RCMs and GCMs to simulate precipitation, and the use of techniques developed in this project to the detection of precipitation changes associated with anthropogenic forcing. The PI and co-PI also have established collaborations with individuals and groups who are well-suited to use the products of this project (such as simulated time series of precipitation to be used to drive streamflow models at CEH) in applications that are more directly beneficial to the general public. The weakest aspect of the proposal is the method proposed to relate SST variability to extreme precipitation. While the rest of the proposal deals well with the climate system processes that relate the large scale atmospheric circulation with precipitation, a direct link between SST variability and extreme precipitation by using metrics of particular SST patterns as covariates in GEV/GPD models appears to miss the dynamics that is likely to be involved in this relationship. Instead, it is more likely that SST will affect atmospheric circulation or temperature, which will then influence precipitation extremes. Thus, I would prefer to see the use of some variety of atmospheric or climate model, or even empirically derived relationships, to first relate SST variability to variability in the atmospheric circulation, which is then related to the distribution of extreme precipitation, rather than looking for a direct SST-precipitation relationship. This is hinted at by the final sentence of the "Method and approach" section, but should be clarified. Otherwise, the approaches proposed are well-suited to the problem of understanding the causes of variability and change in extreme precipitation. I am impressed by the links that the PI and co-PI have made with researchers in other organizations that will be able to provide input to and use the products of this project. The clearest benefit is from the link to NERC Centre for Ecology and Hydrology, which will utilize hydrology model to produce streamflow estimates from simulated precipitation time series to produce information that has more directly applicable benefits than what the PI and co-PI can produce on their own. It also seems that the Hadley Centre collaborators are sufficiently engaged to provide two-way discussions with the PI and co-PI on the direction of this project. It seems that the amount of time that the PI and co-PI are charging to the project is minimal. The expertise of the PI and co-PI does seem critical for the successful completion of this project, so their involvement at least at the level that they are charging to the project appears well-justified, and I expect that they will spend more time on the project than they are charging. In addition, the amount of time budgeted for the project, to be primarily carried out by a postdoctoral scientist, seems reasonable; at the very least, it does not seem to be less time that is needed. -- This message (and any attachments) is for the recipient only. NERC is subject to the Freedom of Information Act 2000 and the contents of this email and any reply you make may be disclosed by NERC unless it is exempt from release under the Act. Any material supplied to NERC may be stored in an electronic records management system.