Influence of non-feedback variations of radiation on the determination of climate feedback

Recent studies have estimated the magnitude of climate feedback based on the correlation between time variations in outgoing radiation flux and sea surface temperature (SST). This study investigates the influence of the natural non-feedback variation (noise) of the flux occurring independently of SST on the determination of climate feedback. The observed global monthly radiation flux is used from the Clouds and the Earth's Radiant Energy System (CERES) for the period 2000-2008. In the observations, the time lag correlation of radiation and SST shows a distorted curve with low statistical significance for shortwave radiation while a significant maximum at zero lag for longwave radiation over the tropics. This observational feature is explained by simulations with an idealized energy balance model where we see that the non-feedback variation plays the most significant role in distorting the curve in the lagged correlation graph, thus obscuring the exact value of climate feedback. We also demonstrate that the climate feedback from the tropical longwave radiation in the CERES data is not significantly affected by the noise. We further estimate the standard deviation of radiative forcings (mainly from the noise) relative to that of the non-radiative forcings, i.e., the noise level from the observations and atmosphere-ocean coupled climate model simulations in the framework of the simple model. The estimated noise levels in both CERES (>13 %) and climate models (11-28 %) are found to be far above the critical level (~5 %) that begins to misrepresent climate feedback.