NCAR / ucomp-pipeline

Data processing pipeline for UCoMP
Other
6 stars 3 forks source link

Explore optional color tables for LOS to avoid drawing the eye from the zero point #278

Open bberkeyU opened 5 months ago

bberkeyU commented 5 months ago

The current LOS color table blue-white-red does an excellent job of drawing the eye to locations in the image where the velocity flips from positive to negative. However, our zero-point uncertainty may give a false impression of where things change in the corona.

Perhaps we should consider a perceptually uniform sequential map like Viridis or Plasma (matplotlib maps). If these can be loaded into idl or a similar idl map, it would be nice to look at some complex coronal configurations, including CMEs, to see if these still pop out and if other events pop out instead.

Tasks

mgalloy commented 5 months ago

Color tables 49-65 are the Brewer sequential color tables that would be appropriate for this. We could also construct a custom one.

mgalloy commented 5 months ago

Here are some of the color tables for the 1074 waves mean LOS velocity image for 20220208.

Color table 49:

20220208 ucomp 1074 l2 waves mean los_velocity-49

Color table 49:

20220208 ucomp 1074 l2 waves mean los_velocity-50

Color table 51:

20220208 ucomp 1074 l2 waves mean los_velocity-51

Color table 52:

20220208 ucomp 1074 l2 waves mean los_velocity-52

Color table 53:

20220208 ucomp 1074 l2 waves mean los_velocity-53

Color table 55:

20220208 ucomp 1074 l2 waves mean los_velocity-55

Color table 65:

20220208 ucomp 1074 l2 waves mean los_velocity-65

bberkeyU commented 5 months ago

I think 49 and 65 look cool and bring out more information in the redshifted regions. My eye has an easier time seeing finer structure and detail in these areas. Plus, for some reason, the icy blueness of 49 seems stunning. And some of the blue-shifted structures near PA230 really pop in 65.

Unfortunately, the whites (blue-shifted regions) in all of these color tables seem to wash together. I can't tell the color differences once the blue shifts get below bigger than -2 km/s. I think I am seeing a change in adjacent structures, but I dont think I can compare the whites from pixels that aren't adjacent they all look white/pale.

Using the Matplotlib plasma color table addresses some of my concerns about the lack of contrast in the white and high-blue shift regions.

I think the plasma table shows finer detail than the old table. I was hoping the new tables would give better quicklook insights into the data, but at least for Feb 08, I think I have looked at the data enough that I am now seeing the same features across all tables. I think more days will need to be reviewed to see if switching tables helped with quick insights.

The biggest drawback I am seeing with the new tables is my deep connection between red/blue and velocity. The other tables dont keep this connection and I have in reviewing the images I keep finding myself needing to refer to the color tables to keep track of the sign of the velocity. If we do adopt one of these tables, I am pretty sure our expert users will quickly adapt, but this seems like it could be a real problem for new users.

image