Understanding Measurements Of Climate Sensitivity Grant uri icon


  • The mean temperature of the earth is determined by the balance of incoming and outgoing radiant energy at the top of the atmosphere, and the outgoing energy generally increases and decreases with mean temperature. The temperature dependence of outgoing energy stabilizes earth''s climate, as an increase in temperature produces an increase in outgoing energy which cools the planet, and a decrease in temperature has the opposite effect. The restoring effect of temperature on outgoing energy can be quantified by a climate feedback parameter which summarizes the net effect of a variety of mechanisms which together determine the global effect. For internally generated climate variability such as El Nino, which produces a temporary increase in global temperature, the feedback parameter determines how quickly the temperature will return to its long-term climatological value. For externally forced climate change, such as the warming produced by an increase in carbon dioxide (CO2), the feedback parameter determines the amount of warming that will ultimately occur as a result of the radiative effect of the CO2 increase.It is commonly assumed that the same feedback parameter applies very generally to global temperature changes produced by internal climate variability (like the temperature increase during El Nino events) and permanent climate change forced by increases in carbon dioxide (CO2) and other external factors. But preliminary work by the PI and others suggests a systematic difference between parameter values calculated from internal variability and forced change simulations in climate models, possibly due to differences in the spatial patterns of temperature anomalies associated with internal variability and forced change, and the fact that different feedback mechanisms are prominent in different regions. A difference in feedback strength between forced response and internal variability would complicate efforts to estimate the feedback parameter from observations, as the necessary observations include satellite measurements of the incoming and outgoing top-of-atmosphere radiative fluxes that are only available for a 15-year period. Changes in radiative fluxes over such a short period are dominated by internal variability, thus the feedback parameter value derived from them would only be representative of internal variability and should not be used to assess the sensitivity of climate to CO2 increases. A further issue identified by the PI is that the feedback parameter may have strong decade-to-decade variability, at least in long model simulations of climate under present-day conditions. Such variability, likely also related to the spatial patterns of temperature anomalies and their consequences for specific feedback mechanisms (for instance the albedo feedback associated with ice and snow which generally occur at higher latitudes), would also have to be taken into account when attempting to estimate climate sensitivity from observations.This project considers possible differences in feedback parameter between internal variability and forced change, as well as the possibility of decadal variability in the feedback parameter, using a combination of observations and model simulations. Much of the work focuses on an alternative feedback parameter based on the mean temperature at a mid-tropospheric level (500mb), as parameter values calculated at this level show greater agreement between observations and simulations than their counterparts based on surface temperature. This agreement motivates further examination of the processes contributing to differences in feedback parameter in model simulations, which have the advantage of very long periods of record and outputs which include detailed breakdowns of hard-to-observe quantities such as the longwave and shortwave radiative effects of clouds.The work has broader impacts due to the desirability of constraints on how much warming can result from increases in atmospheric CO2 and other greenhouse gases, particularly given the large uncertainty in estimates from model simulations. The project also supports two graduate students, thereby providing for the future workforce in this research area.

date/time interval

  • 2017 - 2020