Open question on non-surge, non-tidal components of water levels

[WPringle]

Most of SurgeMIP models just run "storm tide" (astronomical tide + wind and pressure-driven barotropic surge).
So running this detiding operation on those outputs should give a really good signal of "storm surge".

The main issue comes when we compare "storm surge" from observations (GESLA) which contain other components.
There are lower frequency non-barotropic signals due to seasonal heating/cooling, ocean currents, mesoscale eddies, sea-level rise etc. How do we treat the observation when analyzing? Should we run a say a 30-day moving mean filter over them before or after detiding?

In addition, models like the ECCC one does include baroclinic components, so we may need to consider how to fairly treat those model outputs for comparison.

[chaigneaualisee]

I agree this is an important point that we need to address.
As currently implemented (at least from my understanding), the detiding procedure removes all fitted tidal constituents, including the long-period constituents Sa and Ssa (annual and semi-annual). As a result, it already removes part of the low-frequency variability present in both tide-gauge observations and baroclinic model simulations, such as the seasonal cycle, although it does not remove all the low-frequency baroclinic variability. This could affect comparisons between observations, purely barotropic storm-tide simulations, and model configurations that include baroclinic processes.