On a couple of occasions, I’ve noted that near infrared water vapor parameterizations in HITRAN-1996 were incorrect and wondered about what the impact of these changes would have been on a non-retuned GCM. It looks like Collins et al JGR 2006 have done something like that – implementing HITRAN changes up to 2003. Unfortunately, they did not attempt to implement the most up-to-date changes as of 2006 (or even implement 2004 changes). Collins et al 2006 excuse this on the following basis:
In general, the radiative parameterizations in GCMs are updated infrequently and are not consistent with the latest LBL calculations of H2O shortwave absorption.
While this may be standard practice in GCM circles, given the amount of weight being placed on GCM prognostications, this seems like a pretty casual way of doing business, especially given that Collins et al are aware of other changes:
There are independent indications that the near-infrared absorption calculated using the latest HITRAN may be underestimated.
However, that’s not the topic of today’s post. It turns out that the difference between old and new parameterizations is about 3.4 wm-2 and that this amount is absorbed in the atmosphere. Collins et al:
The updates to the parameterization of water vapor extinction increase the shortwave convergence in the atmosphere by approximately 3.4 W m-2.
The amount (3.4 wm-2) is obviously very similar to the forcing from doubled CO2 (3.7-4 wm-2). So in a sense, Collins et al, by testing the impact of the 3.4 wm-2 change in atmospheric “convergence”, may provide interesting information on sensitivity where there are no built-in expectations for the answer. I say this because it’s hard not to think that tests on the impact of doubled CO2 within climate models – from a statistical point of view – have become so stylized that they amount to little more than a type of Kabuki theatre. Here there are no built-in expectations. Before you read on, write down your guess as to the impact on average temperature obtained by Collins et al and compare it to their result which I’ll give below.
I do not claim any particular authority on these models and am posting up some notes on my reading, which I’ve tried to do carefully but I could easily have misunderstood something. So bear that in mind. Collins et al describe the difference in parameterizations as follows:
In this paper, we demonstrate the effects of recent changes in water vapor spectroscopy on the climate simulated with the Community Atmosphere Model (CAM). In the original shortwave parameterization for CAM [Briegleb, 1992], the absorption by water vapor is derived from the LBL calculations by Ramaswamy and Freidenreich . In turn, these LBL calculations are based upon the 1982 AFGL line data [Rothman et al., 1983]. The original parameterization did not include the effects of the water vapor continuum in the visible and near-infrared. In the new version of CAM, the parameterization is based upon HITRAN-2K [Rothman et al., 2003], and it incorporates the CKD 2.4.1 prescription for the continuum. For the broadband shortwave absorption of interest here, the differences between using CKD 2.4.1 and MT_CKD are negligible.
They report the following direct effects of the changes:
The updates to the parameterization of water vapor extinction increase the shortwave convergence in the atmosphere by approximately 3.4 W m-2. … This increase is accompanied by a increase in total solar energy absorbed at the top of the climate by 0.6 Wm-2 and a reduction in surface insolation by 2.8 Wm-2.
They go on to say that the impact of these changes is primarily on the hydrological cycle:
The change in surface insolation is balanced primarily by a reduction in latent heat flux, although the sensible and net upward longwave fluxes also decrease… The global precipitation rate falls by -0.05 mm d-1 to balance the reduction in surface evaporation. The net effect of the greater near-infrared absorption is to weaken the hydrological cycle by approximately 2% while increasing the amount of precipitable water by the same percentage.
Their Table 8 shown below shows the following impacts of these changes on a coupled model – the slab ocean seems more relevant here than the prescribed ocean. As I understand it, the impact of increased atmospheric absorption in the amount of 3.4 wm-2 on a coupled model with slab ocean is 0.122 K. I’ve written to Collins asking him to confirm that my understanding of this is correct (and, if not, to provide what their estimated impact is).
Now I understand that there are differences between an “increase in shortwave convergence” of 3.4 wm-2 and an increase in “radiative forcing” of 3.7 wm-2 due to doubled CO2. I realize that one is inbound radiation and one is outbound radiation; I realize that one is shortwave and one is longwave. I just don’t understand why there is a difference of 1.5 orders of magnitude in temperature impact, by which one leads to a change in temperature of 0.122 K and the other leads to projected changes of 1.5 to 4.5 K, with some models yielding even higher results. I’m not saying that any of the calculations are wrong – merely that I’m puzzled by the seeming disproportion in results. I firmly believe that this sort of effect should be explicable in relatively simple terms without needing to rely on GCMs and would welcome any explanations.
Collins, W. D., J. M. Lee-Taylor, D. P. Edwards, and G. L. Francis (2006), Effects of increased near-infrared absorption by water vapor on the climate system, J. Geophys. Res., 111, D18109, doi:10.1029/2005JD006796.