cc: "Keith Briffa" date: Fri, 08 Apr 2005 14:29:52 +0100 from: Tim Osborn subject: Re: Emailing: ECHO-G and solar weighting to: "Rob Wilson" At 10:36 08/04/2005, Rob Wilson wrote: >1. I have to admit that I still find it strange that the natural forcing >ECHO-G run does not show higher values in the recent period when compared >to the 1100-1225 period. The higher modelled temperatures through this >earlier period are obviously forced by solar input. The fact that the >natural run does not show equivalently high (or higher) values in the >recent period means that the relative weighting of the forcings must have >changed over time. This seems completely illogical to me, but in my >ignorance, I think I must go with what the model shows. However, I think >some sort of mention of this is warranted in the paper - or am I being too >simplistic about all this. I believe the early warmth in the ECHO-G run is due, at least in part, to the initial conditions being too warm (nearer to present-day than pre-industrial) and hence would have been cooler if the model had started with more realistic conditions. At least partly an artefact of the experimental design rather than the real world. I tried to show that at the Reading meeting with comparisons with the MAGICC simulations. >2. As you and Keith will be included on the coral paper (will not hear >back from Sandy until the 20th) we can discuss this issue further later as >this is relevant for that paper also. I am troubled about the variance of >the modelled data being higher than the actual instrumental data. For >example, the temperature amplitude (defined as the difference between the >warmest and coldest decade) in the ECHO-G model of NH temperatures is 2.08 >(see table below). However, the variance of the model is much higher than >the instrumental data over their common period. If I scale the modelled >data to the same mean/variance of the instrumental data (in this case Land >only 20-90N), then the amplitude reduces to 1.27. This value sits nicely >between my RCS reconstruction and Jan's curve when they have also been >scaled to the same instrumental data-set. This analysis suggests that >amplitude of the Moberg series is possibly too high. Also, ECHO-G has no cooling effect in 20th century due to tropospheric sulphate aerosols and hence warms too much. Hence amplitude too great. Tim >Coldest > >Warmest > >Amplitude > >Wilson STD > >1813-1822 (-0.74) > >1938-1947 (0.20) > >0.94 > >Wilson RCS > >1600-1609 (-0.97) > >1937-1946 (0.17) > >1.14 > >Esper02 > >1345-1354 (-1.18) > >1950-1959 (0.15) > >1.34 > >Briffa00 > >1813-1822 (-0.80) > >1951-1960 (0.10) > >0.90 > >Mann99 > >1458-1467 (-0.68) > >1957-1966 (0.10) > >0.79 > >Jones98 > >1693-1702 (-0.77) > >1929-1938 (0.06) > >0.83 > >Moberg05 > >1576-1585 (-1.33) > >1104-1113 (0.23) > >1.56 > >ECHO-G > >1689-1698 (-1.97) > >1953-1962 (0.11) > >2.08 > >ECHO-G [scl] > >1689-1698 (-1.09) > >1953-1962 (0.18) > >1.27 > >Does the higher variance in the ECHO-G model imply that it is TOO sensitive? > >comments are much appreciated > >regards >Rob > > > >----- Original Message ----- >From: Tim Osborn >To: Rob Wilson >Cc: Keith Briffa >Sent: Wednesday, April 06, 2005 10:22 AM >Subject: Re: Emailing: ECHO-G and solar weighting > >At 13:07 05/04/2005, you wrote: > >I was always under the impression that, in general, solar changes > >controlled long term changes in climate and volcanic events caused short > >term cooling. > >... > >I guess 'clusters' of volcanic events could cause a longer term response > >of the climate system. > >Clustering of volcanic events does indeed enable them to have longer term >influences. But there probably is a changeover in which is dominant as you >go to longer time scales. But where that changeover happens depends on how >strong the forcings are, which is particularly uncertain for solar (and >fairly uncertain for volcanic). The strength of the forcings is determined >by how the modellers designed their experiments - they specify the forcing >time series in advance. For the ones they chose to use in ECHO-G (based on >Crowley's earlier work, though not identical) the volcanic still looks to >be dominant on time scales up to 30 years and probably of similar >importance to solar on time scales 30-100 years (these are my guesses from >eyeballing some figures of smoothed forcings, but could be quantified by >looking at spectra etc.). Above 100 years, then solar is probably bigger >than volcanoes in ECHO-G, but not overwhelmingly so. > > >How are the relative weightings of the external forcing estimated in the > >models? > >Although I agree that GHGs are important in the 19th/20th century > >(especially since the 1970s), if the weighting of solar forcing was > >stronger in the models, surely this would diminish the significance of GHGs. > >The forcings for ECHO-G are selected in advance by (1) choosing the >strength and time series of solar irradiance variability; (2) choosing the >strength and time series of volcanic aerosol variability and converting >this to a surrogate time series of solar irradiance reductions, which are >then added to (1); and (3) choosing the time series of greenhouse gas >concentrations. > >Thus (1) and (2) prescribe the forcings to the model - there is no role for >the model itself to determine the strength of those forcings. The model >does however determine the strength of the forcing induced by the GHGs, but >this is fairly accurately known as are the histories of the GHG >concentrations (the combined uncertainty in history of GHGs and their >conversion to radiative forcing is given by the IPCC TAR as just >+-10%). Thus the ECHO-G model's only role in determining the relative >weightings of the forcings is in how it converts (3) to a forcing and this >is very well known. Hence the climate model doesn't really determine the >relative weightings - they are determined by whoever designed the >experiment and selected the inputs. > >If (1) had been chosen with stronger changes, ECHO-G would have responded >by making last 150 years warming stronger - the GHG-induced warming would >still be just as great, but would have extra solar-induced warming on >top. Thus GHG-warming would not be reduced. This might give a poorer fit >to the observed record, implying either that the solar forcing should be >reduced (since the GHG can't be as it's so well known) or that the climate >model is too sensitive (quite possible). So, in the latter respect, your >point could be correct because a lower sensitivity would then downplay the >response (including the GHG-induced warming) until it fitted the >observations once more. > >However, that recent paper by Foukal et al. suggests the evidence for >long-term fluctuations of solar irradiance is biased and perhaps there >isn't any! Also, the people doing detection & attribution of 20th century >climate changes reckon that the particular timing of the solar and GHG >forcings during the 20th century (i.e. the interdecadal time series >structure, not just the increasing trend) and the slightly different >patterns of climate response to the two forcings can be used to distinguish >between the two and they find the observed warming time series and patterns >match better to that expected from the GHG forcing than the solar forcing - >and if I understand it correctly, this result doesn't depend on the >magnitude of the forcings, just the time series structure. > >Sorry to go on. Hope some of this is useful/interesting. > >Cheers > >Tim > >Dr Timothy J Osborn >Climatic Research Unit >School of Environmental Sciences, University of East Anglia >Norwich NR4 7TJ, UK > >e-mail: t.osborn@uea.ac.uk >phone: +44 1603 592089 >fax: +44 1603 507784 >web: http://www.cru.uea.ac.uk/~timo/ >sunclock: >http://www.cru.uea.ac.uk/~timo/sunclock.htm Dr Timothy J Osborn Climatic Research Unit School of Environmental Sciences, University of East Anglia Norwich NR4 7TJ, UK e-mail: t.osborn@uea.ac.uk phone: +44 1603 592089 fax: +44 1603 507784 web: http://www.cru.uea.ac.uk/~timo/ sunclock: http://www.cru.uea.ac.uk/~timo/sunclock.htm