Add McBryde projection of metric maps to nightsum pages

[ehneilsen]

No description provided.

[rhiannonlynne]

The plots look nice in general but I have some questions:

• are the visits from last night getting included into the plot values? The dark circles don’t seem to have anything interior to them.
• are the dark circles useful to indicate visits from last night, when one visit last night = 1 circle and ten visits last night = 1 circle? Is this getting the information across that is needed? I find the dark circles don’t tell me enough about what happened last night but also make it hard to see the background of all visits, which is kind of the opposite of what you’re going for. If both “visits from last night” are the goal and “visits from all time” - maybe it’s worthwhile to have both.
• the red circles don’t seem useful for the McBryde plots. An alternative would be to just center the LST (RA) at the middle of the night for the plot, so the current sky is centered for the reader and leave it at that. The plots are small to begin with, and having a lot of additional lines on top of the content makes it hard to pick up details.
• I realize that the above opinions on the plots are my opinions. Should we try to gather some more people for their viewpoints?

[rhiannonlynne]

There is a reference to “SV” in the summary table (“time available for SV visits”) — maybe should reword to “Night time hours” ?

[ehneilsen]

There is a reference to “SV” in the summary table (“time available for SV visits”) — maybe should reword to “Night time hours” ?

Or perhaps renamed to "LSST visits" or "science visits." It's not quite "night time hours," in that it includes just science visits, not all visits, while many of the other figures include all figures. The list of block ids it selects on to find "science visits" can be adjusted further, if need be.

[ehneilsen]

I realize that the above opinions on the plots are my opinions. Should we try to gather some more people for their viewpoints?

I think it's less of a question of aesthetics than use cases: I had specific uses for these plots when I made them, and when other users have other uses, we need to supply other plots.

In particular, the purpose of the depth map plots that are part of this PR is to answer the question: "Did the scheduler do a good job filling in shallow areas (holes) in the survey?" To do this, you need the black circles (what indicated exposures taken tonight) to not be filled in, because if they are, you can't see the color "behind" them (the coverage previous to tonight), and so can't tell if it's filling in shallow areas of the footprint. Similarly, you need to be able to see whether it was possible to observe any given area on the night, which makes the red lines (at least the dashed part) necessary: to answer the question the plot is intended to answer, the user needs to be able to look at any given area of the sky (that potentially looks shallow) and know whether it was observable give the time of year at all. Yes the position of the sun gives some vague indication of that, as would marking the zenith at sunrise and sunset, but the shape of the edge between "observable" and "unobservable" is complex enough that anything even close to borderline will be hard to guess by looking at the plot. A simple heuristic based on RA, for example, works poorly: you can see this from the fact that the red lines (which show the actual border of what's visible) do not follow RA graticules well at all. For the azimuthal/planisphere plot, I can guess at the border a little bit better that just a line of constant RA (it's just an oval that rotates around the pole), but for the McBryde map with its weird variable half-omega shape, it's not something at least I can really visualize easily from just a zenith or sun position.

Similarly, the time since last visit plot is supposed to indicate how well the scheduler is maintaining cadence on visible parts of the sky. As in the case of the depth map, if the outlines of the pointings from last night are filled in, that defeats the purpose of the map, as does leaving off the edges of what's observable.

In other words, phrased as a response to this question:

If both “visits from last night” are the goal and “visits from all time” - maybe it’s worthwhile to have both.

neither "visits from last night" nor "visits from all time" are goals on their own: the goal is to compare visits from last night to visits from all time spatially.

Of course, "visits from last night" and "visits from all time" are both interesting on their own, as are a number of other spatial quantities. I'm working on SP-2518 now, which adds a number of "just from last night" maps (see the comments of the Jira issue for a list). These are responses to requests made at the RCW. (In some cases, I'm not really sure what questions they're all supposed to answer. For example, one of them is a map of the number of bands observed twice. I can see how something like this would be useful, but for the specific questions I can think of, histograms would better than maps, so I'm not sure.)

"Visits from all time" questions are also vital, but I think these are more appropriate for progress reports instead of night summaries. Work on progress reports has been on the back-burner for a while, but is something I'm eager to work on more. Again, the progress report plots will be made with different design choices than the plots in this PR. The rising and setting airmass/zd limits are certainly not needed for them, for example.

[ehneilsen]

One way I think the depth plot here can be improved greatly is to replace the raw depth with the difference between depth and the variable footprint as it was on the night, as used in the scheduler basis function. I think making this replacement should be a separate PR, though.