Period dependent short-term shortwave and longwave feedback parameters derived from CERES observation

Abstract:

Period dependent shortwave and longwave feedback parameters are derived from the CERES EBAF-TOA data from March 2000 through May 2014. The algorithm uses time series of top-of-atmosphere reflected shortwave and emitted longwave irradiances, as well as surface skin temperature monthly deseasonalized anomalies. The time series is converted to the period domain by Fourier transfer and feedback parameters are derived from the amplitude ratio of the reflected shortwave or emitted longwave anomalies to the surface skin temperature anomalies multiplied by the cosine of the phase shift (FFT approach). While feedback parameters vary significantly depending on the period, they appear to converge as the period increases. Once they are sorted into period bins with the width of 1 year and mean values from all bins are averaged, the global mean value agrees with that derived from a simple linear regression to their uncertainty. The FFT approach provides a method to screen feedback parameters with shorter periods (less than a year) that contribute to the variability significantly. While feedback parameters derived from current CERES observation of 15 years differ from climate feedback parameters, their relationship to climate feedback parameters can be tested with climate models. The advantage of the FFT approach as opposed to a linear regression is that it can derive time-scale dependent feedback parameters. In addition, period dependent feedback parameters can be used to assess a linear system assumption for shorter periods (less than 10 years) and provide a guide on the length of the data record needed to accurately infer climate feedback parameters.