Caldeira and Myrhvold  attempted the obvious by compiling various global warming models to try to extract simple thermal behavioral patterns, see their online paper.
They succeeded in my opinion.
What is most important about their work is that they placed diffusional models of warming, as pioneered by James Hansen , on an equal footing with first-order kinetics model. The first-order models use damped exponentials and are favored by analysts that want to keep the math simple. Caldeira and Myrhvold understand this and provide mixed exponential models that will piece-wise map to the diffusional responses normally seen in the numerical simulations.
The Figure 1 below shows the various responses to a 4× step input of CO2 applied to a steady-state climate. The CO2 generates a greenhouse gas forcing to the earth’s surface, which is partially absorbed by the vast expanse of the ocean acting as a thermal heat sink. The depths of the ocean gradually equilibriate with the surface as the heat diffuses downward — but until the layers reach a steady state, the temperature will continue to creep up. Note the fast transient, the Transient Climate Response (TCR), followed by a gradual increase to the Equilibrium Climate Sensitivity (ECS) .
The foundational methods that I employ in models of thermal diffusion are described in a paper Diffusive Growth (also linked by the menu item at the top of this blog). This model assumes a dispersion of thermal diffusivities and uncertainty in the impulse depth. Figure 2 below provides a general generic view of how this model is applied.
If one compares the shape of the step response of Figure 2 with the responses shown in Figure 1, the overall agreement is apparent. This is not unexpected as diffusional flow is the solution to the heat equation, which is quite general in its applicability. The differences arise mainly in the abstraction needed to describe the effective diffusion of heat in the ocean — this is a combination of various eddy diffusivities at different scales. I describe this also here in terms of a Ocean Heat Content model.
The DCS has a few interactive screens for evaluating the thermal diffusion curves, see http://entroplet.com/context_thermal/navigate for an example.
This diffusional response can be added to the CSALT model of global warming. Consider Caldeira and Myrhvold’s Figure 5 reproduced below as Figure 3, which gradually adds 1% additional CO2 per year. The fast TCR intuitively tracks the growth of CO2 so that it is difficult to distinguish the CO2 growth from the transient response as long as the CO2 accumulation continues. (This is very similar to the Red Queen growth in Bakken reserves described elsewhere, where the diffusional production is difficult to separate from the well growth as long as wells are accelerating).
The full diffusional response will eventually be added to the CSALT model as a convolution response term. Right now the land temperature warming is considered instantaneous as it reaches its full ECS response quickly, while the ocean warming reaches an average TCR representative of the most recent accumulative CO2 warming value (with a slight lag). It is not an absolutely critical addition yet it will add the extra fidelity that a comprehensive model should include, especially in the out-years as the diffusional effects accumulate and the ocean TCR inches closer to the ECS.
 K. Caldeira and N. Myhrvold, “Projections of the pace of warming following an abrupt increase in atmospheric carbon dioxide concentration,” Environmental Research Letters, vol. 8, no. 3, p. 034039, 2013.
 J. Hansen, D. Johnson, A. Lacis, S. Lebedeff, P. Lee, D. Rind, and G. Russell, “Climate impact of increasing atmospheric carbon dioxide,” Science, vol. 213 (4511), pp. 957–966, 1981.