Characterizing Coronal Hole Structure with Feature Tracking Where Feature Tracking Hasn't Gone...

Characterizing Coronal Hole Structure with Feature Tracking

Where Feature Tracking Hasn't Gone Before… (a report on work in progress)

Brian Welsch & Hugh HudsonSpace Sciences Lab, UC-Berkeley

Loren ActonDept. of Physics, Montana State University

After staring at many SXT movies, Loren noticed some- thing about wide-latitude coronal holes, e.g.,

To wit: the visibility of these coronal holes (CHs) is not E-W symmetric across the disk.

In particular, such CHs tend to be “more visible” to the East.

Q: How common is this effect? A: Examples are common in wide-latitude CHs.

It’s hard to see the effect in wide CHs; only “narrow-ish” exhibit the effect.

But this effect is only a tendency – it’s not ubiquitous.

PHOTOSPHERE

CORONA

West Limb East Limb

Along some lines of sight (LOS; gray dotted lines above), more closed-field loops exist, and their higher densities produce more emission along those lines of sight.

A prograde tilt is unexpected, since the Parker Spiral in the outer corona corresponds to a retrograde tilt.

LOS #1 LOS #3LOS #2

Q: What causes this effect? A: Loren’s idea was that CHs tend have a prograde tilt.

PHOTOSPHERE

CORONA

West Limb East Limb

The visibility of untilted coronal holes should be symmetric with respect the E-W angle of differing lines of sight (gray dotted lines), because similar amounts of closed-field loops exist lines of sight from E & W.

Using a “quasi-cartoon”, one can test the symmetric visibilty null-hypothesis.

(Images generated with Google’s SketchUp software.)

How can this effect be quantified? We have focused on CH areas.

• We initially studied E-W intensity variations, but Hugh argued that SXT’s S/N is too low in CHs for such variations to be meaningful.

• So our approach was to define a “reference hole”, then remap it, to study E-W changes in the areas below threshold.

One can then compare the instantaneous CH area with the visible area of the remapped reference CH.

(If not restricted to the reference hole, instanta-neous CHs can extend to other regions.)

(Results are in-sensitive to whether rigid or differential rotation is assumed.)

The instantaneous fraction of pixels below threshold does show E-W asymmetry.

Minor Problem:

‘the’ longitude of CHs is poorly defined – CHs often extend over a wide E-W range.

Major Problem: The choice of frame used to define the reference CH affects the curve of E-W fractional area.

In this test using the synthetic data, a frame with the primary CH east of disk center was used.

This produced an asymmetric E-W fractional area curve.

Changing the reference frame changes the shape of the curve.

So I scrapped defining reference holes, and started identifying segments of holes – a.k.a., features.

• The low-latitude disk was divided into strips.

(The polar holes are too wide, so were excluded.)

• A dual-threshold, gradient-based algorithm identified hole segments:

Pixels had to be both below a high threshold and strictly uphill from a low-threshold pixel.

(Simple thresholding selected either too few or too many pixels.)

Tracking the identified hole segments from frame to frame was challenging, for several reasons.

• Intensity fluctuations between frames caused some features to disappear temporarily!

• Changing labels with merging/ fragmentation should probably be avoided.

Sadly, however, I’ve still got some tracking bugs!

This might be a failure of association (from my artificially slow cadence?), or new feature label bookkeeping, or ???

Aside: Loren’s original characterization was that CHs got “fuzzier” from E W.Extract fix-latitude intensity cuts at successive times, then take dI/dx.

Intensity contours.

East-edge “trough” fadesAcross disk.

Aside (cont’d): One can also discern the decrease in CH sharpness in grayscale images.

Left: const.-lat. intensity cuts show “the” CH.

Right: grayscale of dI/dx shows that CH edges are less sharp on the Western disk.

Red intensity contours help match features between the plots.

Solid white lines mark the solar limb.

Edges fadenear W limb.

Edges sharpnear E limb.

What might cause this prograde tilt?

• Net tilt of photospheric field? (Welsch)– Shrauner & Scherrer (1994): 0.5o prograde tilt of

photospheric field (cf., 4.5o tilt in bipoles)

• Back-reaction from solar wind acceleration? (Hugh)– Admittedly crackpot, but the numbers are ~plausible.

• Typical AR morphology – compact leading field, diffuse following field

West Limb

Since the leading polarity of AR bipoles tends to be more compact, the flux density (and hence magnetic pressure) is greater on the preceding side of ARs than on the following side.

As the simple potential field model shown above illustrates, open field (black field lines) at the eastern/ following edges of coronal holes therefore has a prograde tilt.

We have noted some tilt in PFSS models of CHs that we’ve tracked.

East Limb

Future Directions

• Automate this tracking technique, and analyze the entire Yohkoh database – determine freq. of tilted CHs, & solar cycle variability

• Look for this effect in other satellites’ data…

– STEREO+EIT would be ideal for this study, but the effect isn’t obvious in EUV

– Inspection of randomly sampled XRT movies doesn’t reveal the effect there, either. (But XRT has too many filter settings!)

Conclusions• Loren noted an interesting effect: coronal

holes appear to possess a prograde tilt.

• We’re attempting to quantify this effect in SXT data.

• We have some ideas about what might cause this tilt.

• Our work is still very much in progress…