cc: t.osborn@uea, k.briffa@uea, p.jones@uea date: Fri Sep 18 10:51:20 1998 from: Tim Osborn subject: GR3/12107 response to referee's comments to: kec@wpo.nerc.ac.uk Dear Karen On behalf of Keith Briffa, Phil Jones and myself, I have attached our responses to the referee's comments regarding our NERC research grant proposal (GR3/12107). They are included as a Word 7.0 document attached to the e-mail, and repeated below as plain text. Please acknowledge receipt of this message, and confirm that you can read our comments. Regards Tim -------------------------------------------------------- 16 September 1998 Dear Ms Campbell Thank you for the opportunity to respond to the referees’ comments on our invited resubmission (GR3/12107; ML:21). In general, the various remarks could justifiably stimulate protracted discussion, though we have restricted ourselves to relatively brief responses. These are ordered as in your letter: Referee A Many of this referee’s remarks are most appropriately viewed as comments rather than ‘problems’ or questions. We wish to say that, in almost all respects, we concur with the spirit of what is said. We agree that "in much recent work .... there has been an over-reliance on apparently sophisticated analysis techniques" which make it "difficult to evaluate the significance .... of any relationships observed". However, the referee is mistaken (perhaps because of the way we described our intended methodology) that we consider such methods to be the "techniques of first resort". On the contrary, we argue strongly (in our published work and in this proposal) that simpler techniques are always to be preferred: ones where explicit attention can be given to quantifying the nature of the association between palaeo and real/modelled data. This is not to say that the more complex techniques we specify are not useful. Indeed they can represent very efficient and objective methods for summarising and comparing the associations between the various data sets - provided (as the referee implies) that there exist direct, local relationships between the palaeodata and the relevant primary observed (or modelled) climate parameter, one that can be expressed in the form of a simple linear equation (with some autoregressive error term). Incidentally, the more sophisticated spatio/temporal techniques need not necessarily provide ‘better’ estimates of the (regression) model parameters, if they provide a more efficient or easily demonstrable means by which the patterns of interest may be portrayed. However, they should provide reconstructions that are demonstrably no worse than simpler techniques. We wish to investigate this and issues of comparative reconstruction confidence in this proposed work. We do reassure the referee and the committee, that our analyses were never intended to apply the ‘sophisticated’ techniques exclusively and we had always intended to include the ‘explicit modelling’ referred to by the referee. We wish to make one final point, as regards referee A’s remarks. We believe the simple model associations (and hence reconstruction uncertainties) between palaeodata and the observed data are frequency-dependent [we discuss this in Jones et al. (1998) referred to in the proposal] and require appropriate quantification in explicitly specified period bands, hence the suggested use of some of the techniques we list. Again, this issue has not been adequately addressed in the recent literature. In summary, we disagree with nothing in principle said by this referee and indeed their remarks suggest that, were this proposal to be successful, some collaboration would prove very beneficial to its outcome. Referee D The proposal is ambitious - perhaps somewhat overly ambitious - but this is an important and extensive area for research. Given the collaboration with others that we hope for, however, we are determined to make a valuable contribution. It is somewhat contradictory of this referee to then suggest we adopt a more global perspective for this research. This would significantly increase the workload. We made a pragmatic decision to concentrate on the Northern Hemisphere (or near Northern Hemisphere - we intend to include tropical data). Extra-tropical Southern Hemisphere data are not as numerous or as well documented as data north of 20°S. True, the mechanisms that drive climate change may operate outside of the region upon which we are focussing. However, that focus is only on issues of palaeoquality and the scope and veracity of climate inference; and the relationships with model data. The GCM model experiments are global in extent. It is not our intention to deal with model deficiencies - merely to identify what these are, or might be, on the evidence of the palaeodata. It is indeed true that the recent NAO trend is outside the range of "natural" variability defined by some GCM simulations (Osborn et al., 1998). As the referee then states, the question is, are the models deficient or are the recent trends "unnatural"? It is precisely for this reason that we wish to reconstruct an alternative estimate of natural variability of the NAO (and other indices and modes), based on proxy data. We note also that models with anthropogenic forcing or forced by observed SST (which has part of the anthropogenic signal in it) are able to reproduce part of the recent observed trend in the NAO (Osborn et al., 1998; Mark Rodwell and Chris Folland, pers. comm., 1998). Referee G 1. True, most of the available NAO-related palaeodata display stronger coherence with the observations in summer. The Moroccan tree-ring-based reconstruction of winter precipitation will, therefore, be important, but it is not our only source of data. Besides existing long instrumental data (e.g. CET) we will have significant new and longer ‘instrumental’ data (e.g. Cadiz MSLP data to 1776) because of other work at CRU. This will include an extended NAO record, itself extending back 200 years. Other proxies include ice core winter isotopes (Barlow et al., 1993), west Greenland ice accumulation (Appenzeller et al., 1998), tree-ring chronologies with some sensitivity to winter temperatures (Cook et al., 1998) and others. 2. Yes it does suggest that either the spatial pattern of response (as reflected in different climate parameters) has changed - or that the response pattern has remained the same, but some other spatial signal has degraded the NAO signal. This is a complex problem but one that can only be addressed by maintaining separate, pressure-related (including instrumental) evidence of the NAO variability and independent proxy evidence of the response pattern. We wish to explore this issue in collaboration with Dr. J. Hurrell, who we have recently learned will be visiting the Hadley Centre for an extended visit during 1999. 3. Page 3, paragraph 2, line 12 was poorly expressed. Not all the proxies go back 1000 years and the CET record only extends to 1659. Referee H All of this research, including part (i), will be done in close collaboration with modellers at the UK Met. Office and at DKRZ in Hamburg. Indeed, Drs. Simon Tett and Ulrich Cubasch will be actively collaborating with us and providing direct input to the project, and we are in close contact with other modelling/palaeo institutions that are now proposing similar investigations (e.g. Dr. Hans von Storch at GKSS in Germany and Dr. Nanne Weber at KNMI in Netherlands). We hope that the above points clarify the issues raised by the referees. If the committee requires any other clarification, we would be happy to attempt to respond. Yours sincerely Keith R. Briffa Phil D. Jones Timothy J. Osborn