Velocity based mechanism to explain small scale contribution

[jbusecke]

Past Friday we talked through a simple model which might explain the small scale contribution we diagnose from the model run.

@rabernat asked me for a histogram of the velocity magnitude of the unsmoothed and smoothed input velocities (relative velocities).

I made a temporary plot (only for ~2 year of the data):

I am unsure how you diagnose a 'typical' change in velocity from here, but happy to make further changes to the plot.

[rabernat]

Hmm that certainly doesn't look like a very big change. 🤔 There is a very slight shift in the distribution to lower values. Can we reconcile that with the results from #47?

[jbusecke]

Yeah I expected more too, but this is also a very large range of velocities!
I wonder if a change of e.g. 0.25m/s (I want to say that could be seen at the y-value of ~1?) would suffice to explain the change we are seeing?
Is there a metric (maybe the mean or something related to the skewness) that I could compute to quantify this better?

[dhruvbalwada]

Seems odd, but maybe possible. Is this looking like this because most of the relative velocity is coming from atmosphere wind, which is not impacted much by filtering?

• Do you have maps of snapshots of the filtered and un-filtered velocities?
• Did you smooth velocities and then compute magnitudes or smooth velocity magnitudes?
• Is this plot a histogram before or after time average?
• Can also look at histogram of anomalies : Unfiltered - filtered velocities.

[rabernat]

• Can also look at histogram of anomalies : Unfiltered - filtered velocities.

☝️ this sounds like a promising approach. Would probably have to make it | Unfiltered - filtered velocities |.