Title: A coarse-grained decomposition of surface geostrophic kinetic energy in the global ocean

Abstract: We apply a coarse-grained decomposition of the ocean's surface geostrophic flow derived from satellite and numerical model products. In the extra-tropics we find that roughly $60\%$ of the global surface geostrophic kinetic energy is at scales between $100~$km and $500~$km, peaking at $\approx300~$km. Our analysis also reveals a clear seasonality in the kinetic energy with a spring peak. We show that traditional mean-fluctuation (or Reynolds) decomposition is unable to robustly disentangle length-scales since the time mean flow consists of a significant contribution (greater than $50 \%$) from scales $<500~$km. By coarse-graining in both space and time, we find that every length-scale evolves over a wide range of time-scales. Consequently, a running time-average of any duration reduces the energy content of all length-scales, including those larger than $1000~$km, and is not effective at removing length-scales smaller than $300~$km. By contrasting our spatio-temporal analysis of numerical model and satellite products, we show that the AVISO gridded product suppresses temporal variations less than 10 days for all length-scales, especially between 100 km and 500 km.