Characterization Observing Conditions

This family shows the observing conditions at every attempted characterization of an Earth-like planet -- working angle, delta magnitude, phase, and SNR -- and which of those attempts failed. The question it answers is not "how many planets did the mission characterize" but "what did the hard ones look like, and is there a pattern to the failures."

The data source is reduce-earth-char-list.csv.gz, produced by the yield reduction in reduce_drms.py. Plots are made by plot_drm_earth_chars.py, are named det-earth-char-wa-dmag-*.png and det-earth-char-hist-phi*.png, and appear in the Characterization Observing Plots section of the ensemble page.

The planet class is the configurable "earthlike" class, so a scenario that redefines it in config-reduce.json -- to a Sub-Neptune population, say -- gets that name in the plot titles instead.

The data

One row per planet per characterization observation. A row is an attempt, not a planet: an Earth-like planet whose star is characterized three times contributes three rows, and a star with two Earth-like planets contributes two rows per observation. Counts on these plots are therefore counts of attempts, and do not match the yield counts elsewhere on the page.

The columns used are the observing conditions WA (mas), dMag, phi, char_SNR, and MV (the star's V magnitude), plus four flags: is_success, n_success, is_deep, and is_promo.

Phi is the Lambertian phase function, between 0 and 1. It is already folded into dMag; a planet at an unfavorable phase is dimmer by 2.5 log₁₀ Phi magnitudes, which is why Phi gets two plots of its own.

Outcomes

Every row falls in exactly one of three outcomes, and the distinction between the last two matters for reading the scatter plots:

Condition	Outcome	Meaning
`is_success > 0`	success	this planet was characterized on this observation, in some band
`is_success == 0`, `n_success > 0`	split	this planet failed, but another Earth-like planet at the same star succeeded on the same observation
otherwise	failure	nothing was characterized

Splits are not a curiosity: a multi-planet system is characterized as a system, so whenever an observation succeeds for one planet and not another, every planet in it lands in this middle category.

Target populations

Two flags describe the star, and they are made exclusive before plotting:

deep-dive -- is_deep: the star was on the deep-dive list.
promoted -- is_promo and not is_deep: promoted to the characterization list, with deep-dive taking precedence.

Only the promoted population gets scatter plots -- the deep-dive panels were retired in 2024 -- which is why every WA/dMag filename carries -promo-. The histograms stack both populations.

If the scenario has no deep-dive targets, the blue series is empty and its legend entry reads "(N/A)". That is the common case, not a failure.

The "easy" box

Every scatter plot draws a red dashed rectangle at WA > 70 mas and dMag < 26, and an inset counts the successes and failures inside it. These bounds are fixed constants, not read from the scenario's instrument parameters, so on a mission whose inner working angle or contrast floor differs markedly the box is a fixed reference mark rather than a statement about feasibility. The "failed" count in the inset covers only the plotted (promoted) population; the "successful" count covers all of them.

The scatter plots

All four share a layout: working angle on x, delta magnitude on y, with the y-axis clipped at zero and the x-axis left to autoscale.

Grey + markers are every successful characterization in the ensemble, across all populations. They are context -- the cloud the mission actually managed.
Colored dots are the unsuccessful promoted attempts, shaded by the quantity the plot is named for.

One asymmetry is worth knowing. The SNR plot separates the two bad outcomes, drawing failures as dots and splits as x markers, and its legend counts them apart. The phi and vmag plots merge them: their colored dots are every attempt that was not a success, splits included. The same plot region can therefore look denser on the phi plot than on the SNR plot.

The shadings:

snr -- char_SNR, the SNR actually achieved. Low values at small WA and large dMag are the expected shape.
phi -- log₁₀ Phi, on a color scale pinned to [-2, 0], so everything at Phi < 0.01 saturates at the bottom. Rows with Phi exactly 0 land there too.
vmag -- the planet's apparent visual magnitude, MV + dMag. This is the one plot that depends on how bright the host star is.
zzz-dmag (extra) -- the y coordinate is a phase-adjusted dMag, showing where each planet would sit if it were observed at a reference phase of Phi = 0.7. Planets already above that phase are not moved, and the correction is capped at 5 magnitudes. Comparing it against the plain phi plot separates "too dim to characterize" from "caught at a bad phase." The zzz in the name only sorts it last on the page.

The phase histograms

Both histogram only the failed characterizations, stacked by population (blue deep-dive below, orange promoted above).

hist-phi bins Phi linearly, 25 bins across [0, 1]. A second y-axis on the right carries a line per population giving the failure probability in each bin -- failures over attempts at that phase. The bars say where the failures are; the lines say whether that phase is actually hostile or merely well-populated. Read the lines where the bars are tall enough to support them.
hist-phi-log bins the same failures by 2.5 log₁₀ Phi, the magnitude penalty the phase imposes, in 0.4-magnitude bins from -6 to 0. The leftmost bar is a catch-all for everything below -6, including the Phi = 0 rows. There are no rate lines on this one.

The plots

Five are always produced; one more with mode_op set to +.

Plot	Content	Extra?
`wa-dmag-promo-snr`	WA/dMag, shaded by achieved char SNR; splits drawn separately
`wa-dmag-promo-phi`	WA/dMag, shaded by log Phi, scale pinned to [-2, 0]
`wa-dmag-promo-vmag`	WA/dMag, shaded by planet apparent V magnitude (`MV + dMag`)
`hist-phi`	failed chars vs. Phi, with per-bin failure probability
`hist-phi-log`	failed chars vs. 2.5 log Phi, the phase magnitude penalty
`wa-dmag-promo-zzz-dmag`	WA/phase-adjusted-dMag, shaded by Phi	x

To get the extra plot for every family, make S=... graphics-extra. To ask for it for this family alone -- and have it stick, rather than depending on which target was last run -- add this to the config-reduce.json of the scenario, or of the family or experiment above it:


  "graphics": {
    "mode_op": {
      "earth_chars": "+"
    }
  }

earth_chars is this family's name in the plot registry; plot_drm_driver.py --list shows them all.

Caveats

These are attempts, not planets. Nothing here is a yield. A frequently revisited star dominates the point count.
Successes and failures are drawn from different populations. The grey successes span the whole ensemble; the colored failures are promoted targets only. Do not read a success:failure ratio off the plot.
char_SNR is the achieved SNR, not the threshold. The plots do not draw the scenario's SNR requirement, so where the success boundary should fall is not marked.
A planet can appear as both. Success is per observation, so the same planet can be a grey + from one visit and a colored failure from another.