Status Update:
Activity has decreased to ~7% after reaching an all-time high of
~ 17% (flux lost due to dust clouds, averaged over an orbit).
The "high phase" set of several dips continue to march to higher
phases. My favorite dip, Dip#1, is fading to almost nothing; it
has lasted at least 6.7 months (and probably 8.4 months). At
least one dip drifts with a period the same as the A asteroid,
and it may have 3 or 4 companions drifting with the same period
(all other dips have very slow drift rates with respect to the A
asteroid fragments). It seems to have disappeared a few weeks
ago.
Figure 1a. Latest phase-fold LC.
Figure 1b. Previous good-quality phase-folded LC.
Notice the monotonic trend for fading within the phase region
+0.11 to +0.25.
The phase region showing fades in the previous figure does
indeed show a simultaneous change at all phase values within
that region starting ~ Jan 21. [Needs update]
Figure 2. Activity level vs.
date. Activity level is defined as the sum of "area under
the curve" for all dips occurring during a 4.5-hour orbit.
A 45-fold increase is apparent during parts of the 8
months of our amateur team observations during the 2015/16
observing season compared with the Kepler K2 observations
2 years earlier. This increase may have occurred after a
hypothetical collision between a fragment and the "mother
ship" asteroid in 2015 September (Event 1). Activity level
declined steadily starting in 2016 April, and by the end
of the first observing season (2015/16) activity had
returned to a level almost as low as before the September
2015 event. However, the first few observations of the new
observing season (starting 2016 Oct) shows an uptick of
activity, reaching ~ 80 times the K2 level of activity (as
of 2017 January). The rise in activity leveled off, and is
now decreasing.
Figure 3. Expanding the date scale shows when event #5
may have occurred. It had to be between 2016.07.23
(the last measurement of the 2015/16 observing season)
and 2016.11.06 (when it was first noticed, at a level of ~
4%). The model shows event #5 with a "box" to
indicate possible suggested onset dates (i.e., for the first
group of dips). Event #6 has a
well-established date of 2016.11.19 (the first green
symbol).
Waterfall diagram for 2016/17 season. Dip #1 has lasted
175 days (5.8 months). Dip "A" has the same period as what we
used to refer to as the "mother asteroid" so maybe the "A"
period is really for L2 fragments. That would make the bulk of
dips coming from L1 fragments, with only occasional in-between
dip periods associated with the asteroid "mother ship."
There appears to be a collision, or break-up, of the fragment
producing Dip#1 in mid-February, leading to 3 to 5 smaller
fragments that produce smaller and occasional dips.
Another display format with objective solutions for drift
line fitting. Adopted period corresponds to my estimate of the
"A" asteroid (4.495 hr). Drift lines sloping to the left
(going up in time) have periods shorter than the ephemeris
period (my adopted "A" asteroid); drift lines sloping to the
right have longer periods; drift lines with no slope (moving
straight up) would have the same period as my adopted "A"
asteroid. Most drift lines slope to the left because more
fragments drift away from the L1 end of the asteroid because
it is hot; very few drift lines slope to the right because
very few fragments drift away from the L2 end because it is
cool (perpetual shadow). Drift line 1 has lasted 137 days (4.5
months). [Needs to be updated]
Properties of Drift Line #1 (cf. previous figure).
It's fading, and currently has the lowest depth since it was
first detected (2016 Nov 6). [Needs updating]
Most fragments are expected to come from the L1 end of the
asteroid, so most dust clouds should have the L1 fragment
periods. Fewer fragments should come from the L2 end because
it is cooler (because it's in shadow). I don't know if the
asteroid itself can produce dust clouds because of its
gravity.
Links on this web page:
Some
physical model thoughts
List
of observing sessions
Observing
session observations
Finder image
Data
exchange files
References &
related links
My
collaboration policy
Some Physical Model
Thoughts and Appeal for "Sweet Spot" Telescope Observations
Figure 3 in the previous section shows an "activity slope"
of 3.7 %/day starting with Event #6. Let's consider this to be
characteristic of how fast a dust cloud's "total projected area"
increases
with time after a "collision" event. The term
"total projected area" refers to the projected area of dust
particles with circumferences larger than optical wavelengths
(i.e., Mie scattering regime). Imagine that a "collision" event
produces particles with a size distribution that ranges from
microns to meters, for example. The small particles will be
pushed outward from the WD by radiation pressure, so they will
be in orbits with ever-longer periods, causing the dust cloud to
expand in phase to larger values (i.e., a trailing tail). This
is what we see for the second dip group in Fig. 4, above.
Presumably, all particles will produce molecules of minerals due
to sublimation, and these may recondense to Mie size small
particles. The smallest source particles will evaporate first
(due to sublimation), and thereby be removed from the original
particle size distribution. The remaining larger particles will
sublimate, and become smaller. It should be possible for
detailed models of these processes to make use of the 3.7 %/day
activity slope as an observational constraint.
We can also imagine that the first group of dips underwent this
process starting at an earlier date (as suggested by Andrew
Vanderburg, private communication). Note that the first dip
group has a trailing tail appearance. Note further that this
group has a stable "lead-off dip" that is sharp and stable in
phase; this could be a dust cloud produced from a population of
large particles whose orbits haven't been affected by radiation
pressure.
Note that I haven't taken a position on whether the asteroid and
its fragments are consolidated rock or rubble piles (like almost
all solar system asteroids smaller than a certain size). If the
parent body asteroid is a rubble pile, the "fragments" that
break away (from the L1 end) would start out as a "clump of
rubble," possibly held together initially by inter-molecular
forces (in the same way that asteroids small enough to violate
the 2.2-hour "spin barrier" are held together). These, in turn
will be destined to come apart, especially if they rotate fast;
fresh surfaces would then be exposed to heating on all sides.
Sublimation will erode outer surfaces, and the strong
gravitational field gradient will flex the rubble clump, causing
it to expand further. The lowest density rubble clumps will
break apart before the denser ones (i.e., at larger distances
from the WD).
In this model we are dealing with fragments composed of "gravel"
with an inherent size distribution. For fragments that are
actually a "clump of rubble" instead of a solid rock it is
easier to produce "events" of increased activity. This is
because we no longer require actual "collisions" to initiate
dust cloud production. Instead, all we require is that a "clump
of rubble" be disturbed enough by the various forces acting on
it to begin a process of coming apart. The coming apart will
release "gravel" with a size distribution encompassing a wide
range of sizes. The smallest particles will have the greatest
surface area per mass ratio, and can be expected to dominate
dust cloud production immediately after "clump of rubble"
breakup. All particles will present "fresh surfaces" for
sublimation. The argument about radiation pressure producing a
trailing tail can be invoked. When a "clump of rubble" fragment
begins to break apart it may produce chunks of rubble that
slowly separate, turning the pile into a "rubble cloud." Less
shading within a cloud will lead to more sublimation (Chiang,
private communication). If the rubble clump breaks apart from
spinning the smaller clumps, or rubble clouds, will separate
from each other. Each such rubble cloud will produce dust clouds
due to sublimation, and when these are of a size to block a few
percent of WD light in their orbit the dips produced will form
drift lines in a waterfall diagram that have slightly different
slopes. Assuming the dust clouds last for several weeks, as they
have been noted to do, the fade pattern will consist of several
dips that spread apart week after week. Projection of the
waterfall drift lines backward to a convergence date would allow
determination of the rubble clump's break apart date.
The above rubble pile "model" predicts that multi-wavelength
measurements of dip depth will show that the trailing side of a
group exhibits greater amounts of wavelength dependence of depth
vs. wavelength (i.e., consists of a smaller population of dust).
We need 2- or 3-color simultaneous LCs during the two dip
groups, G6901 and G6B18.
Arguments like the ones above will possibly be used by modelers
when there is sufficient observational evidence to work with.
We need better photometric monitoring than I can provide with an
amateur 14-inch telescope. I'm limited to 1-minute exposures
with SNR that is only sufficient for characterizing 5% dips that
last for 10 minutes or longer. Brief observations with large
telescopes reveal the presence of 1% dip structure with
10-second temporal variations. There's a "sweet spot" for
telescope size that will allow good SNR while permitting full
orbit (4.5-hour) observing sessions at weekly intervals. I think
that sweet spot aperture is 1 to 2 meters. This is an appeal for
someone to take the lead in obtaining telescope time with a 1 to
2-meter telescope for weekly 5-hour observing sessions for the
duration of the 2016/17 observing season (now to June).