%0 journal article
%@ 1463-5003
%A Stopa, J., Semedo, A., Staneva, J., Dobrynin, M., Behrens, A., Lemos, G.

%D 2019
%J Ocean Modelling

%P 18-29 
%R doi:10.1016/j.ocemod.2018.12.002
%T A sampling technique to compare climate simulations with sparse satellite observations: Performance evaluation of a CMIP5 EC-Earth forced dynamical wave climate ensemble with altimeter observations
%U https://doi.org/10.1016/j.ocemod.2018.12.002
 
%X Global climate simulations do not capture the exact time history, making it difficult to directly compare them with observations. In this study we simulate the sampling of altimeter observations from a seven-member wind and wave climate ensemble. This allows us to assess the skill of the climate simulations, relative to satellite observations instead of the typical approach which uses reanalysis or hindcast datasets as reference. Out of the sampling methods tested, we find that a systematic sampling technique performs the best. We then apply systematic sampling to wind fields from EC-Earth and wave fields generated using the wave model (WAM) to replicate the changing sampling of the satellite observations. Next we then quantitatively assess the climate simulations and find that the probability density functions (PDFs) computed from the EC-Earth wind speed samples match the shape of the PDFs obtained from the altimeter observations. EC-Earth consistently underestimates the wind speed with respect to the altimeter observations. Contrary to the wind speed underestimation, the wave simulations overestimate wave heights especially in the extra-tropics. The wind speed seasonality in EC-Earth is larger than the seasonality evaluated from altimeter wind observations while the opposite is true for the wave height seasonality; suggesting the wave physical parameterizations can be improved. We find that the wave height inter-annual variability of the modeled data is considerably less than the inter-annual variability evaluated from the altimeter observations; suggesting long-term climate variability is not well captured. Overall the wave ensemble captures the important features of the global wave climate. The methodology can be adapted to other climate simulations and observational datasets.