## GREB Decomposition: Long Wave Radiation

Diving once again into Conceptual understanding of climate change with a globally

resolved energy balance model (Dommenget and Floter, 2011), this time to flesh out the emissivity and long wave radiation within the model.

3.2 Thermal radiation

The thermal radiation forcing to T_{surf}is due to the black body emission of the surface and due to the atmospheric downward thermal radiation, which depends on the atmospheric temperature, T_{atmos}, the CO2 concentration, the vertical integrated atmospheric water vapor concentration, viwv_{atmos}, and the cloud cover. It is the only way the greenhouse gas CO2 influences the climate system, but it also represents the most fundamental negative feedback to increasing T_{surf}and the most important positive feedback due to atmospheric water vapor response. Simple climate models (e.g. Harvey and Schneider 1985; Weaver et al. 2001) use different approaches to simulate the long wave radiation, which mostly include some parameterizations of the surface outgoing and atmospheric downward radiation (e.g. Ramanathan 1977; Ramanathan et al. 1979; Fanning and Weaver 1996). We try to keep the model as simple as possible with the least number of parameters and base our thermal radiation model on the slab atmosphere greenhouse model (Bohren and Clothiaux 2006).

Two take away points here. First OLR depends on T_{atmos}, CO2, water vapor, and cloud cover. Second, this is a slab atmosphere greenhouse model.

The net thermal radiation is due to a loss by outgoing thermal radiation and a gain by atmospheric thermal radiation:

F

_{thermal}= -sigma*T^{4}_{surf}+ e_{atmos}*sigma*T^{4}_{atmos-rad}The atmosphere is radiating with the temperature T

_{atmos-rad}and an effective emissivity, eatmos. T_{atmos-rad}will be defined in the context of the atmospheric temperature, T_{atmos}, in Sect. 3.4. The effective emissivity, e_{atmos}, is depending on the CO2 concentration, the vertical integrated atmospheric water vapor concentration, viwv_{atmos}, and the cloud cover. As a simple approximation of the dependency on CO2 and viwv_{atmos}we use a log-function approach as in Myhre et al. (1998). We also need to consider that the absorption and emission of thermal radiation at different spectral bands are overlapping for CO2 and viwv_{atmos}(e.g. Kiehl and Ramanathan 1982). This can be approximated by:e

_{0}=

pe_{4}* log [pe_{1}* CO2^{topo}+ pe_{2}* viwv_{atmos}+ pe_{3}]

+ pe_{5}* log [pe_{1}* CO2^{topo}+ pe_{3}]

+ pe_{6}* log [pe_{2}* viwv_{atmos}+ pe_{3}]e

_{0}is the emissivity without considering clouds first. CO2^{topo}is the atmospheric concentration of CO2 scaled by changes in surface pressure due to the topographic height, z_{topo}: CO2^{topo}= e^{-ztopo/zatmos}* CO2. The scaling height z_{topo}= 8,400 m is a measure of the thickness of the atmosphere. The first term RHS (right hand side) in Eq. 4 is the emissivity due to CO2, viwv_{atmos}and some residual component, pe_{3}, in spectral bands where the components thermal emissivity overlaps. The second and third term RHS are the non-overlapping spectral bands CO2 and viwv_{atmos}terms (they still overlap with pe_{3}). The parameters pe_{4–6}give the relative importance of each absorption band, pe1_{3}are the greenhouse gas species scaling concentration, which we assume to be the same for each absorption band to simplify the approximation. The log-function approach is a simple approximation to consider the saturation effective. Note that this e-function can be 1, but in all simulations discussed in this study such values are not reached. It needs to be noted that the slab greenhouse is an approximation. If greenhouse gasses increase, a multi layer model would be better, which effectively means that e_{atmos}can become larger than one (Bohren and Clothiaux 2006). If no greenhouse gasses exist e_{atmos}should be zero. It is an extreme case (e.g. no CO2, VIWV and cloud cover), but should be considered.

Two main take aways here as well. First is the thermal forcing component defined in this equation:

F_{thermal} = -sigma*T^{4}_{surf} + e_{atmos}*sigma*T^{4}_{atmos-rad}

The second is the emissivity calculation which is divided into three parts: a CO2 component, a water vapor component, and a mixed CO2-water vapor component. Each component is weighted by the values p_{4,5,6}. The CO2 and water vapor terms are scaled by pe_{1} and pe_{2} respectively. Emissivity due to “other” green house gasses (pe_{3})is distributed equally through the three components. This 3 bin treatment of emissivity is designed to separate those portions of the spectrum which is opaque to CO2 from those that H2O opacity dominates from those with significant overlap in the two. See the figure below. This model doesn’t try to calculate the opacity line by line but uses a simple weighting of these bins instead.

Both the previous equations, as well as the following, can be identified in the ‘LWradiation’ subroutine shown below.

There is one more piece to this puzzle – cloud cover.

Cloud cover is considered by:

e

_{atmos}= ( (pe_{8}– CLD) / pe_{9}) * (e_{0}– pe_{10}) + pe_{10}Thus cloud cover, CLD, scales the effective emissivity, e

_{atmos}, by shifting it up or down and by diluting the effects of the trace gasses. So in the presence of clouds emissivity is larger and the effect of trace gas concentrations is reduced. This also reduces the sensitivity to changes in trace gasses concentration. The parameters of the model are fitted to literature values constraining the effective emissivity as function of CO2, viwv_{atmos}, and CLD with an iterative numerical fitting routine minimizing a cost-function (see Appendix 2 for details).

My final take-away here is that all of these equations are heavily parameterized.

!+++++++++++++++++++++++++++++++++++++++ subroutine LWradiation(Tsurf, Tair, q, CO2, LWsurf, LWair_up, LWair_down, em) !+++++++++++++++++++++++++++++++++++++++ ! new approach with LW atmos USE mo_numerics, ONLY: xdim, ydim USE mo_physics, ONLY: sig, eps, qclim, cldclim, z_topo, jday, ityr, & & r_qviwv, z_air, z_vapor, dTrad, p_emi, log_exp ! declare temporary fields real, dimension(xdim,ydim) :: Tsurf, Tair, q, LWsurf, LWair, e_co2, e_cloud, & & LWair_up, LWair_down, e_vapor, em e_co2 = exp(-z_topo/z_air)*CO2 ! CO2 e_vapor = exp(-z_topo/z_air)*r_qviwv*q ! water vapor e_cloud = cldclim(:,:,ityr) ! clouds if(log_exp == 11) e_vapor = exp(-z_topo/z_air)*r_qviwv*qclim(:,:,ityr) ! sens. exp. linear-function ! total em = p_emi(4)*log( p_emi(1)*e_co2 +p_emi(2)*e_vapor +p_emi(3) ) +p_emi(7) & & +p_emi(5)*log( p_emi(1)*e_co2 +p_emi(3) ) & & +p_emi(6)*log( p_emi(2)*e_vapor +p_emi(3) ) em = (p_emi(8)-e_cloud)/p_emi(9)*(em-p_emi(10))+p_emi(10) if(log_exp == 11) em = em +0.022/(0.15*24.)*r_qviwv*(q-qclim(:,:,ityr)) ! sens. exp. linear-function LWsurf = -sig*Tsurf**4 LWair_down = -em*sig*(Tair+dTrad(:,:,ityr))**4 LWair_up = LWair_down end subroutine LWradiation

So grabbing the emissivity knob firmly, it is easy to twist emissivity up to 1 and down to 0 as seen in the chart above. Dommenget and Floter often cite Bohren and Clothiaux 2006 which is an atmospheric radiation text. Their back-of-the envelope for e=1 is a T_{e} of 303K and for e=0 is T_{e} of 255K. (I have only sampled the B&C text and am not sure T_{e} is a surface temperature.) GREB runs a little cooler with an e=1 leading T_{surf} of 293K which is a bit cooler than B&C’s 303K. But the real divergence is on the low end, where an e=0 leads to a T_{surf} of 238K. This is quite a bit lower than most “no greenhouse” estimates I have seen and is a cautionary note regarding GREB’s emissivity model and parameterization.

—————–

Dommenget, D., and J. Floeter 2011: Conceptual Understanding of Climate Change with a Globally Resolved Energy Balance Model. Climate dynamics, 2011, 37, 2143-2165.

Hi, I think your website might be having browser compatibility issues.

When I look at your blog site in Opera, it looks fine but when opening in Internet Explorer, it

has some overlapping. I just wanted to give you a quick

heads up! Other then that, fantastic blog!