date: Mon, 23 Jan 2006 09:31:50 +0000 from: simon.tett@metoffice.gov.uk subject: [Fwd: Clim. Dyn] to: Richard Betts , Tom Crowley , Jonathan Gregory , Tim Johns , Andy Jones , Elisabeth Ostrom , Tim Osborn , Margaret Woodage All, attached are reviewers comments on our paper. All are fairly positive and minor. In addition I need to include some information on land-sfc changes (t/s of forest area) and correct the soil moisture text. [I will post a copy of the review to David Roberts] Simon -- Dr Simon Tett Managing Scientist, Data development and applications. Met Office Hadley Centre (Reading Unit) Meteorology Building, University of Reading Reading RG6 6BB Tel: +44 (0)118 378 5614 Fax +44 (0)118 378 5615 Mobile: +44-(0)77 538 80696 I work in Exeter about 2 days/week. E-mail: simon.tett@metoffice.gov.uk http://www.metoffice.gov.uk Received: from exxmail1.desktop.frd.metoffice.com ([151.170.112.5]) by exxmail2.desktop.frd.metoffice.com with Microsoft SMTPSVC(5.0.2195.6713); Fri, 20 Jan 2006 14:36:25 +0000 Received: from exxmgw1-fw.metoffice.gov.uk ([151.170.243.15]) by exxmail1.desktop.frd.metoffice.com with Microsoft SMTPSVC(5.0.2195.6713); Fri, 20 Jan 2006 14:36:24 +0000 Received: from hermes.cnrs-gif.fr ([157.136.10.1]) by exxmgw1-fw.metoffice.gov.uk with ESMTP; 20 Jan 2006 14:36:24 +0000 X-SBRS: 5.5 X-BrightmailFiltered: true X-Brightmail-Tracker: AAAAAQAAA+k= X-IronPort-AV: i="4.01,205,1136160000"; d="scan'208,217"; a="8193760:sNHT33864584" Received: from mailhost.lsce.cnrs-gif.fr (mercure.lsce.cnrs-gif.fr [157.136.14.3]) by hermes.cnrs-gif.fr (8.12.10/8.12.8) with ESMTP id k0KEaDDR025530 for ; Fri, 20 Jan 2006 15:36:13 +0100 Received: from [157.136.14.188] (tahiti.lsce.cnrs-gif.fr [157.136.14.188]) by mailhost.lsce.cnrs-gif.fr (8.12.11/jtpda-5.4) with ESMTP id k0KEaAKZ004498 for ; Fri, 20 Jan 2006 15:36:12 +0100 Mime-Version: 1.0 X-Sender: duplessy@mailhost.lsce.cnrs-gif.fr Message-Id: Date: Fri, 20 Jan 2006 15:37:48 +0100 To: Simon Tett From: Jean-Claude Duplessy Subject: Clim. Dyn Content-Type: multipart/alternative; boundary="============_-1074354622==_ma============" X-Virus-Scanned: by amavisd-milter (http://amavis.org/) Return-Path: Jean-Claude.Duplessy@lsce.cnrs-gif.fr X-OriginalArrivalTime: 20 Jan 2006 14:36:24.0981 (UTC) FILETIME=[E3D9D450:01C61DCE] Dear Simon, I have now received 3 reviews of your manuscript "The impact of natural and anthropogenic forcings on climate and hydrology since 1550" submitted by Tett et al I will be able to accept it for ublication in Climate Dynamics if you can revise it following the referees comments. If you feel able to follow these recommendations, I will need only 2 paper copies of the manuscript plus a letter explaining changes that you have made. In this cas, i will make the last review myself. I will need together with the anuscript the usual copyrighht transfer form from your institution. With my best wishes for the New Year, Jean Claude Ref 1 Comments on the manuscript "The impact of natural and anthropogenic forcings on climate and hydrology since 1550" submitted by Tett et al. to Climate Dynamics November 2005 The manuscript describes some results of two climate simulations of the past 500 years, one driven by natural forcings only and one driven by natural plus anthropogenic forcings. Both are compared to a control simulation with constant forcing set, for most of the forcing factors, at preindustrial conditions. The manuscript consists of two parts. The first part is devoted to the question of the climate sensitivity, its dependence on the underlying "mean"climate state and on the nature of the external forcing and on the previous climate history. By diagnosing the forcing at the Top of the Atmosphere, the heat fluxes into the ocean and the temperature response of the model, the authors conclude that the concept of climate sensitivity may be dangerous, since in these simulations it depends on all mentioned factors. This result is in itself quite relevant, since it alerts on the dangers of trying to estimate climate sensitive from reconstructions and past temperatures and past forcings and extrapolate this knowledge to future climates. The second part contains a general description of the effects of the natural and anthropogenic forcing on some selected climatic aspects. The description is focused on the effect on variability and on long-term trends. the main conclusion is that the effects anthropogenic forcing is clearly distinguishable from the effects of natural forcings in many important aspects, although not in all of them. In my opinion, quite relevant is the finding that the tropical regions present the largest signal-to-noise ratio, and that detection of the anthropogenic influence on climate could be more successful in this regions, if longer instrumental records or climate reconstructions were available. In this second part, the authors have chosen the strategy of presenting in a quite descriptive way a general overview of the simulations, instead of concentrating in some key aspects and analyze mechanisms, causes and responses in a more detailed way. This is is a matter of taste, but some readers may find this section too descriptive. I think this strategy is alright if the reader is warned at some point that the manuscript has been designed in this way, and that a more detailed process analysis will eventually be published at a later stage. The authors have also chosen not to perform a comparison with reconstructions of global or regional climates of the past centuries. I also agree that the uncertainties in the reconstructions are still too large to warrant a useful discussion on the differences between model and reconstructions. This would also be better placed in a manuscript Regarding the presentation, I think the manuscript is generally in good shape. The relevant references have been included and the figures are in general clear and informative. I have some particular comments that the authors may perhaps want to consider. 1)In page 6, left column, second paragraph, the imbalance of the radiation budget is discussed. Both simulations are found to be in equilibrium with the external forcing until the end of the 19th century. At the end of the 20th century the imbalance due to anthropogenic forcing is 0.2 W/m2. This figures contrast somewhat with those provided by Gregory et al (J. Climate, 2002), who estimate a disequilibrium of the real climate in the 19th century of about 0.2 W/m2 and about 0.32+-0.15 W/m2 in the last decades of the 20th century. Perhaps the authors could refer to these results and comment on their agreement and disagreement. My impression is that the second figures agree, whereas the the first (imbalance in the 19th century) do not. 2) In page 6, the value of alpha (inverse of the sensitivity) and its stability is discussed. The authors find that alpha is not constant along the simulations and suggest that this may be due to the different forcing that come into play along time. However, alpha will also change just because the mean temperature is changing. In a simple energy balance equation the external forcing is proportional to the fourth power of mean temperature. The derivative will be then proportional to the third power of temperature. This would make alpha increase with mean temperature, irrespective of the changing nature of the forcing. The changes in alpha in the all250 simulation go in opposite direction to this expected thermodynamic effect, so that the qualitative effect of the nature of the forcing should be still larger. If changes in the sensitivity due to the nature of the forcing (solar to greenhouse gas) are in the order of 300 %, this result is really a blow to the attempts to estimate the present climate sensitivity from paleoreconstructions. Also in this paragraph, I am not sure that the sentence "although the low correlation will reduce the regression slope" is correct. If , for instance, other sources of temperature variability affect the correlation with the forcing, these sources will increase the temperature covariance with the forcing and the temperature variance in the same amount, so that the regression itself will be unaffected. 3) In page 7, left column, bottom, the effect of orbital forcing is estimated by calculating the trend in the difference between the NH mean from the global mean in the simulation Natural500. However, a trend in the difference may also be present in the control simulation. 4) In page 10, left column, middle, the variance of the simulated Central England Temperature is discussed. Changes in the variability relative to control are possibly ascribed to the coarse model resolution. I think this can be quite correct, since the control simulation has the same resolution. 5)Page 10, right column, bottom. JJA NAO shows a positive trend. This would indicate a pressure decrease in Iceland relative to Azores, and not an increase. 6) Page 11, left column, second half. In the all250 simulation, the aerosol forcing compensates to some extent the greenhouse warming on northern hemisphere land areas through the indirect effect, i.e. the formation of clouds. However, precipitation is reduced precisely on those areas, due to the increased long-wave radiation to space in the troposphere. I think both mechanisms are to some extent opposing each other, unless it is assumed that rain efficiency is strongly changing. Is this point perhaps worth a comment? . 7) The SOI index in figure 11 is defined as Darwin minus Tahiti, which opposite to the usual definition. Is this just a glitch in the caption or has the SOI really been calculated in this way? This could be important because there still exists a disagreement, or uncertainty, about the response of ENSO to changes in the external forcing (Collins, Climate Dynamics, 2005). The analysis of reconstructions and simulations could be, therefore, a key issue in determining the sign of the ENSO response to forcing. Perhaps the authors would to add some discussion on this, at least stating what is the sign of the SOI response to past forcing. If figure 11 has been plotted correctly, with the usual definition of the SOI, it seems that the model would tend to produce a La Niña-like response to increased forcing. If the figure caption is correct, then the model response is more El Niño-like. Some other minor points: 8) the references should be brought to a common format. 9) Page 6, left column, bottom, I think Q should read F. The notation here is different to the one used in Gregory et al. (ibid). Perhaps it can be convenient to use the same notation: Q=F + alpha T 10) When discussing the radiative forcing in all250, the authors refer in many places to CO2 , but methane is not negligible. 11) Page 6, right column, discussion on alpha: some units are missing in these paragraphs. 12) Page 7, right column, middle : change "foring" to forcing. 13) Page 7, left column, bottom: "seas-ice" to sea-ice 14) Page 8, right column: "expected be" to "expected to be" 15) Page 9, right column, middle :"these indices to how stable..." to "these indices to see how stable..." Ref 2 Review of "The impact of Natural and Anthropogenic Forcings on Climate and Hydrology since 1550" by Tett et al., submitted to Climate Dynamics. This paper presents analyses of simulations of the HadCM3 model. The authors emphasize the influence of natural and anthropogenic forcings on climate variability during the past 500 years. The paper is well written and presents a thorough analysis of the variance of temperature and precipitation. I was a bit disappointed that the authors only allude (i.e. use too many synonyms of "suggest" and other hypothetical forms) to mechanisms which could be responsible to changes in variance, and do not offer ways to prove (or disprove) quantitatively such mechanisms in the HadCM3 model. I think that they had the material to be more conclusive. Although it is subliminal, the authors should state in the discussion that their results are model dependent. I have a number of minor comments listed below: 1. Introduction. "one driven by natural forcings from 1492 to 2000" The abstract says 1999. Please make sure that numbers are consistent throughout the text. 2. The authors claim that they cannot compare their simulations with proxy data. They could at least make comparisons with observations since 1850 and check whether the variability they simulate is realistic (and not just for Central England Temperature). 3. Model description, last para. "No attempt is made to model gains or losses from the rivers. Nor do we consider the large anthropogenic changes in river flow through dams, irrigation and sequestration from rivers. Thus our river flow results should be seen as indicative of the possible impact that natural and anthropogenic forcings (including changes in land-surface properties) could have on river flows." Those three sentences mean that the authors acknowledge that the surface hydrological cycle of their model is wrong, and then that ("/thus/") they can retrieve some truth on the hydrological cycle. This sounds illogical and demands a clarification (a true conclusion cannot be drawn from a false assumption). 4. Why chose the year 1492 (apart from historical reasons that do not have much to do with climate)? 5. Comparison with control (bottom of p. 4). Figure 1d does not show river flows. Please check that the figure numbers and captions are OK. 6. Climate sensitivity (p. 6). "As the simulated response is rarely in equilibrium this means that the climate response depends on the rate at which the ocean takes up heat as well as the equilibrium climate sensitivity. Thus palaeo-reconstructions of the last 500 to 1000 years are unlikely, in the absence of other evidence, to provide robust estimates of climate sensitivity." The logical implication ("/thus"/) is not clear to me. Moreover, the purpose of paleo-reconstruction is to obtain temperature (or precipitation) estimates, not to provide estimates of climate sensitivity, especially if "climate sensitivity" is defined by the rate of change under a doubling of CO2. 7. Analysis of climate indices. The authors perform a low pass filter on the climate indices. Could they explain why? 8. Cryosphere (last paragraph). It is very disappointing to conclude a study where all elements are available in principle by sentences including "we think", "this suggests", etc. 9. Regional temperature variability (p. 11). Please explain why less snow cover imply lower temperature variance? 10. p. 11. "Our results strongly suggest that decadal tropical variability is significantly driven by natural forcings. Thus studies into it, and its teleconnections, should consider natural and, as we shall see below, anthropogenic forcings as an important driver of decadal tropical climate variability." I challenge anyone to find tools proving that "decadal tropical variability (DTV) is */not/* driven by natural forcings". The second sentence is even more mysterious to me because it boils down to "not only DTV is driven by natural forcings, but also by anthropogenic forcings". This is uninformative. Apart from those points, I think that it is a valuable paper and I recommend its publication in /Climate Dynamics/ if the proper improvements and clarifications are implemented. Ref 3 Review Tett et al. The impact of Natural and Anthropogenic forcings on climate and hydrology since 1550 General This paper describes the results of three experiments performed with the HadCM3 coupled GCM. Two of these experiments are new and are forced with time-varying natural (experiment NATURAL500) and natural+anthropogenic (experiment ALL250) forcings for the last few centuries. The results of these simulations are compared with a long control simulation with constant forcings, with the aim to study the impact of natural and anthropogenic forcings on climate. This study can be viewed as a follow up of the study by Tett et al. (2002, hereafter T2002), who used the same model to simulate the response to natural forcings from 1850 to 1999 AD. The present study presents a simulation that starts much earlier (in 1492). The aim of this paper is to explore the impact of the various forcings on decadal-centennial scale variability. Unfortunately, the results of NATURAL500 and ALL250 had to be corrected for two sources of error. First, the control experiment experienced a cooling drift, for which both other experiments were corrected using a 2nd order polynomial fit. Second, the set-up of land surface parameters in NATURAL500 contained an error, for which this experiment was corrected using an additional shorter simulation with a correct set-up. These corrections introduce uncertainties that somewhat weaken this paper. In addition, the combination of various forcings in the two experiments makes it difficult to reach conclusions on the impact of individual forcings. Nevertheless, the authors try to do this (e.g., orbital forcing p. 7, deforestation p. 9), but I do find these parts of the paper somewhat unsatisfactory. The experimental set-up does not allow for a quantification of the separate impacts of the forcings, as this would require additional experiments in which the individual forcings are modif! ied. Despite these critical remarks, I find this an interesting paper that discusses relevant new results. Particularly, the conclusion that climate sensitivity depends on the nature of the forcing could have important implications, for instance that simple models are not suitable for climate change studies. Another relevant result is the increase in decadal variance (by a factor 3) in an experiment forced with natural forcings compared to a control experiment with constant forcings. I would therefore recommend publication of this paper in Climate Dynamics after moderate revisions. My suggestions for revision are listed below. Major comments 1) T2002 used an ensemble of simulations, while in this paper Tett et al. have only performed two experiments. Generally, ensembles are performed to get an idea of the variability in the results, which is necessary if one wants to assess the statistical significance of anomalies. Presumably, Tett et al. have not performed an ensemble of simulations to avoid high computing costs. Instead, they have carried out statistical analyses using bootstrap algorithms to assess the uncertainty due to variability and the statistical significance of the simulated anomalies. After reading the present paper, it is not clear to me what the advantages and disadvantages of this statistical analysis are compared to the use of ensembles. It appears that the applied statistical analysis cannot provide the same information as the ensembles. For instance, in ALL250 an early 20th century acceleration is simulated in AMOC (page 9, Section 5.4), but it remains unclear if this due to internal variabili! ty or to external forcing. I therefore suggest that the authors discuss in Section 2 in more detail the differences between the two approaches. 2) In my view, the set-up of CONTROL is not consistent compared to the other two simulations, as it is forced with forcings from varying periods. Greenhouse gas concentrations are from late 19th century, the land surface properties and orbital configuration correspond to 1990 conditions and volcanic aerosol values are the means for the 20th century. It would have made much more sense to use a control experiment forced with forcings for 1750 AD. The use of forcings from different periods makes it problematic to use CONTROL to look at the impact of natural forcings. The NATURAL500 simulation is forced with time-varying forcings for the period 1492 to 1999. Tett et al. analyse the impact of natural forcings by taking the difference between NATURAL500 and CONTROL, and by comparing this relative to the 1700-1749 average from NATURAL500. Because the set of forcings in CONTROL does not represent any period used in NATURAL500, it is not clear what the difference actually represents.! This issue should be discussed in the paper. 3) Page 3, Section 2.2, last paragraph. The authors mention here a drift in the CONTROL simulation and assume that the same drift is present in the other two simulations. I wonder if they tested on what factors this drift depends. Has the sensitivity of this drift to various forcings been tested? For instance, is it known that the drift is the same in a relatively cool (e.g., preindustrial) and a relatively warm climate (e.g., 2*CO2 or 4*CO2)? This would be important to know, also because the uncertainty due to the model drift is not assessed (page 4). Please discuss this issue briefly. Minor comments a) Page 1, Section 1, first paragraph. To support this they used an ensemble of simulations driven by natural forcings alone. This sentence is somewhat confusing, because it appears as if they refers to Collins et al. 2002, who did not use an ensemble, while presumably it refers to T2002. b) Page 2, Section 2.2, first paragraph. Please give the values of the greenhouse gas concentrations used in CONTROL. Are these the same as used for the initial conditions of NATURAL500 (page 3)? c) Page 4, Section3, second paragraph. Over most of the world, temperatures are lower in NATURAL500 than in CONTROL, while in some areas the temperatures are significantly higher. According to the authors, the higher temperatures over continents are associated with reforestation. However, significantly warmer conditions are also found over the central North Atlantic Ocean and over the Arctic Ocean north of Greenland and Svalbard. It is not clear what is causing these relatively high temperatures. Please provide a discussion. d) Page 4, Section 3, third paragraph. Please include a reference to Figure 1b here. e) Page 4, Section 3, fourth and fifth paragraph. Please swap Figures 1c (River flow) and 1d (Soil Moisture), as, according to the text, Figure 1c should show soil moisture and Figure 1d river flow. The same is true for Figures 14c-d, and 15c-d. f) Page 4, discussion Table 1. Please explain in the caption what GS stands for in NH20 Lnd GS Temp. g) Page 5, Section 3. Discussion of taiga-tundra feedback. I suggest including references to Otterman et al. (1984, J Clim Appl Meteorol 23, 762-767) and Foley et al. (1994, Nature 371, 52-54) here. h) Page 5, 2nd column, third paragraph. Total natural and anthropogenic forcing increases at a rate roughly three times faster . I propose to make clear that this sentence refers to the global mean. i) Page 6, discussion of Figure 4. á in ALL250 is not constant (Fig. 4) with very different values of á prior to and post 1900. For 1750-1900 á is 1.8 ± 0.55 W/(m2K) . This is somewhat confusing, as Figure 4 does not show values of á. Instead, the reader is expected to calculate á from the figure. j) Page 8, 2nd column, 2nd paragraph. one might expect changes (omit be). k) Page 9, discussion of Figure 11. I would suggest omitting Figure 11, as it does not provide any additional information. l) Page 10, 1st column, 3rd paragraph. According to Table 2, annual AMOC variability is not significantly different in ALL250. This is inconsistent with discussion on Page 10. m) Page 10, 2nd column, 5th paragraph. Please rephrase sentence The two UK indices show .. n) Page 10, 2nd column, 6th paragraph. JJA NAO in both experiments shows significant positive trend Please discuss the cause of the trend. o) Page 11, 2nd column, line 6. Tropics are warming (instead of is). p) Page 11, 2nd column, 3rd paragraph. These increases in snow cover (instead of increase). q) Figure 12. Please improve labels in Figure 12, these are hard to read. r) Figure 13b. Some information is missing in caption (last sentence). -- Jean-Claude DUPLESSY laboratoire des Sciences du Climat et de l'Environnement Bâtiment 12 Parc du CNRS F - 91198 Gif sur Yvette cedex - tel (33) 01 69 82 35 26 - fax (33) 01 69 82 35 68 - e-mail : Jean-Claude.Duplessy@lsce.cnrs-gif.fr