Quite about IR vs microwave and a nice explanation why this matters.
1000hPa winds because they show force driving the ice more clearly (I think).
As explained earlier, here you want nullschool's 1000hPa 'instantaneous wind power density' WPD map, wind
3 is what moves the floes of ice (freeboard ~turbine blade), not gentle zephyrs of wind
1. They are both derived primarily from air pressure reanalysis/forecast, not measured in situ to any extent.
I've noticed no one has been posting time series of nullschool, presumably because it's a total nuisance layering up screenshots: the slightest mouse movement throws off the nullschool alignment (it has browser page-back but it could be several).
We would further like the nullschool images to be at at the same map scale, projection, and land mass orientation as the WorldView satellite time series, the scalable but non-rotating EPSG 3143, NSIDC's polar stereographic Greenland-down 70º main parallel.
There are two easy ways to script out an animated nullschool overlay of WorldView at maximal forum-compatible size of 700x700.
The first trick is to carefully precondition the view in nullschool's AE projection (azimuthal equidistant). This projection defaults exactly to a north pole-centered -45º rotation relative to rigid WorldView which surprisingly
aligns indistinguishably to EPSG 3143 on the scale of the Arctic Ocean.
Mousing over the screen (never clicking) enlarges the scale
without introducing a rotation, best done on a blank advanced date that has 'no data'. Note the url actually changes in real time as you mouse which can be co-calibrated with WorldView to match scales.
The second trick is to put the mouse down and use keyboard commands instead (below) to cycle through the date range build layers in Gimp (respectively stacks in ImageJ) via whole-window screenshots (command option shift 4 space on a Mac), switching apps back and forth (with command tab/tilde).
Although it takes only ~3-4 minutes for a 10-day time series, including the chained rotation, cropping, a round or two of contrast enhancement, pause-building, frame rate testing and uploading to forum, it pays to turn on keystroke recording to add other nullschool views like temperature.
One frame of nullschool tracers suffices to disambiguate wind direction; there's a keyboard command to turn them off. As noted, the palette has a linear decomposition to grayscale as the L layer of L.a.b color space. This allows replacement of the original ugly palette by arbitrary contoured indexed color which in turn enables quantitation or GIS layer arithmetic interaction.
The second approach starts with an easy tear-down of the starting date's dynamic url into a flat file database of its building blocks. No scripting: build additional dates and incremental views with 'fill-down' spreadsheet commands (kept as template), the new urls are layered with 'Open Location' in Gimp.
Repeating with the same dates in Worldview, the two animations can be butted side-by-side with 'filmstrip' and 'guide-cropping'. In the case of wind WPD, the suspected dominant contributor to daily variation in floe displacement, it makes sense to stagger the times, ie to anticipate WorldView dated movement. The former are furnished at 3 hour intervals (8x per day); the latter are wedge-shaped tiles from different orbital passes.
I will be offline traveling for the next few weeks and will post some examples later