Revisiting the surface-energy-flux perspective on the sensitivity of global precipitation to climate change

Abstract

Climate models simulate an increase in global precipitation at a rate of approximately 1–3% per Kelvin of global surface warming. This change is often interpreted through the lens of the atmospheric energy budget, in which the increase in global precipitation is mostly offset by an increase in net radiative cooling. Other studies have provided different interpretations from the perspective of the surface, where evaporation represents the turbulent transfer of latent heat to the atmosphere. Expanding on this surface perspective, here we derive a version of the Penman–Monteith equation that allows the change in ocean evaporation to be partitioned into a thermodynamic response to surface warming, and additional diagnostic contributions from changes in surface radiation, ocean heat uptake, and boundary-layer dynamics/relative humidity. In this framework, temperature is found to be the primary control on the rate of increase in global precipitation within model simulations of greenhouse gas warming, while the contributions from changes in surface radiation and ocean heat uptake are found to be secondary. The temperature contribution also dominates the spatial pattern of global evaporation change, leading to the largest fractional increases at high latitudes. In the surface energy budget, the thermodynamic increase in evaporation comes at the expense of the sensible heat flux, while radiative changes cause the sensible heat flux to increase. These tendencies on the sensible heat flux partly offset each other, resulting in a relatively small change in the global mean, and contributing to an impression that global precipitation is radiatively constrained.

Appendices

Appendix 1: Calculating the contributions to evaporation change in Eq. (14)

The terms in Eq. (14) were calculated as follows: LE and $R_s$ were taken directly from model output; G was determined from $R_s$, LE, and H based on the surface energy budget (Eq. 4); $\eta$ was calculated from the two-meter air temperature using Eqs. (11) and (13). Finally, given LE, $\eta$, $R_s$, and G, we then solved for $\kappa$ in Eq. (10). The contributions were calculated from ensemble-mean output over the last 5 years of the simulation period. In the equilibrium warming simulations, this was typically 21–25 years after $\hbox {CO}_2$ doubling. In the transient warming simulations, we used years 96–100 after $\hbox {CO}_2$ quadrupling. The contributions were first calculated for each month, and then the monthly contributions were averaged to arrive at an annual-mean value. However, the results were essentially unchanged when the contributions were calculated from annual-mean output.

To understand the global impact of the fractional contributions in Fig. 1, we must account for spatial variability in the magnitude of the mean-state evaporation and surface-air warming. To do so, we multiply each term in Eq. (14) by the following (dimensionless) weighting function,

$$\begin{aligned} w=\frac{E\varDelta T_a}{{\overline{E}}\varDelta \overline{ T_a}}, \end{aligned}$$

(26)

where the overbars in the denominator indicate the ocean-mean values of each variable. These results are then averaged in space, yielding the ocean-mean contributions given in the top left of each panel in Fig. 1.

Appendix 2: Estimating $R_s-G+\kappa$ in the idealized simulations of O’Gorman and Schneider (2008)

To estimate the value of $R_s-G+\kappa$ in O’Gorman and Schneider (2008) simulations, we use the fact that their control climate exhibits a global-mean surface-air temperature of $\overline{T_a}=288$ K, and a global-mean precipitation of 4.3 mm/day, which equates to $L{\overline{E}}=124$$\hbox {Wm}^{-2}$. Given $\eta \approx 0.63$ at $\overline{T_a}= 288$ K, this implies a combined value of $R_s-G+\kappa =197$$\hbox {Wm}^{-2}$. If we assume that this sum is constant, global precipitation is directly proportional to $\eta$, resulting in the gray curve in Fig. 4.