welsch prom06

Progress Report: Tracking Magnetic Features Near PILs: 

Progress Report: Tracking Magnetic Features Near PILs Brian Welsch & Yan Li Space Sciences Lab, UC - Berkeley

Context: This work focuses on observed flows near PILs, particularly shearing and converging flows.: 

Context: This work focuses on observed flows near PILs, particularly shearing and converging flows. Modelers: If not emergence, then either shearing or convergence+cancellation could form prominences. Question: What quantitative constraints do observations place on the amount of shearing and/or cancellation that occurs near PILs? Prominences lead to CMEs. But what processes lead to prominences?

Our goal: to use magnetic tracking techniques to quantify shear and convergence in a sample of active regions (ARs). : 

Our goal: to use magnetic tracking techniques to quantify shear and convergence in a sample of active regions (ARs). Shearing/convergence has been studied before, in cases of 1-2 ARs (e.g., Li et al. 2004); we are studying N > 60. We employ local correlation tracking (LCT) and feature tracking (FT), both described below, to quantify motions. We then attempt to quantitatively characterize the “strength” of shearing and/or convergence. Here, I discuss FT applied to a few ARs: AR 7981 + two reappearances; AR 8100 + one reappearance; AR 8038

We tracked AR 7981’s first appearance on the solar disk…: 

We tracked AR 7981’s first appearance on the solar disk…

… and its second…: 

… and its second…

… and its third.: 

… and its third.

We also tracked AR 8100…: 

We also tracked AR 8100…

…and its next appearance.: 

…and its next appearance.

We also tracked AR 8038, source region for a SHINE Campaign Event CME. : 

We also tracked AR 8038, source region for a SHINE Campaign Event CME.

Data: Our data set includes ARs with and without prominences and/or CMEs. : 

Data: Our data set includes ARs with and without prominences and/or CMEs. To investigate whether “strong” shearing and/or convergence + cancellation are related to prominences and/or CMEs, we must first determine “baseline” flow properties. We can then talk about “3” cancellation rates, etc. So far: we’ve only included ARs from 1996-1998. Data from other years will be tackled soon... So far, our ARs are “bipolar-ish” – i.e., not complex.

Procedure: We deproject the central 45o of full-disk MDI magnetograms, co-align them, then track. : 

Procedure: We deproject the central 45o of full-disk MDI magnetograms, co-align them, then track. We deproject via the Mercator projection, which is conformal (preserves shape locally), but non-authalic (does not preserve area). We approximate Br ~ Blos/cos. Only pixels with |Br| > 20G are tracked. Tracked displacements are scaled to remove distortion.

Velocity inversions generate a 2D map v(x1,x2)from one 2D image, f1(x1,x2), to another, f2(x1,x2). : 

Velocity inversions generate a 2D map v(x1,x2) from one 2D image, f1(x1,x2), to another, f2(x1,x2). The map depends upon: the difference f(x1,x2) = f2(x1,x2) – f1(x1,x2) assumption(s) relating v(x1,x2) to f/t, e.g.: continuity equation, f/t + t(vtf) = 0, or advection equation, f/t + (vtt)f = 0, etc. Based on the assumption chosen, v(x1,x2) is not necessarily velocity – e.g., group velocity of interference patterns.

Local correlation tracking (LCT) finds v(x1,x2) by correlating subregions; it assumes advection.: 

Local correlation tracking (LCT) finds v(x1,x2) by correlating subregions; it assumes advection. 1) for ea. (xi, yi) above |B|threshold… 2) apply Gaussian mask at (xi, yi) … 3) truncate and cross-correlate… * 4) v(xi, yi) is inter- polated max. of correlation funct = = =

From Jan. – Jun. 2006, we implemented an automated LCT “pipeline,” that analyzed current MDI FD m’grams.: 

From Jan. – Jun. 2006, we implemented an automated LCT “pipeline,” that analyzed current MDI FD m’grams. http://solarmuri.ssl.berkeley.edu/~welsch/public/data/Pipeline/ cron checks for new magnetograms with wget New magnetograms are downloaded, deprojected, and tracked. The output stream – including deprojected m’grams, flows (.png graphics files & ASCII data files), and tracking parameters – is uploaded to our web server. These automated procedures enabled us to analyze many more ARs than we originally thought we could!

Once tracked, one can define “the” PIL, to decompose flows into shearing (parallel) and converging (perpendicular) parts. : 

Once tracked, one can define “the” PIL, to decompose flows into shearing (parallel) and converging (perpendicular) parts.

Preliminary results from our LCT study are on-line, at http://sprg.ssl.berkeley.edu/~yanli/lct/, including: 

Preliminary results from our LCT study are on-line, at http://sprg.ssl.berkeley.edu/~yanli/lct/, including full event list, with terse event descriptors; mpeg movies of FD and cropped AR magnetogram time sequences we’ve tracked; fits of PILs

This (LCT + PIL fit) technique works with simple, “straight” PILs; otherwise, it’s problematic.: 

This (LCT + PIL fit) technique works with simple, “straight” PILs; otherwise, it’s problematic. To move beyond this (LCT + PIL fit) approach, we have begun using feature tracking (FT), too.

Unlike LCT, which returns v(x1,x2), FT follows identified “features.”: 

Unlike LCT, which returns v(x1,x2), FT follows identified “features.” (cf., Eulerian vs. Lagrangian fluid descriptions.)

Features are identified with a “ranked down-hill” labeling algorithm (D. Longcope’s idea). : 

Features are identified with a “ranked down-hill” labeling algorithm (D. Longcope’s idea). “Hilltop” pixels are grouped: a pixel takes the label of the “hilltop” to which the field intensity gradient points.

To track, the initial image is segmented into features…: 

To track, the initial image is segmented into features…

…as is the second.: 

…as is the second.

Next, features in the more recent time step are re-labeled, via overlap, to take on the older labels. : 

Next, features in the more recent time step are re-labeled, via overlap, to take on the older labels.

Each feature’s location, flux, size (pixels), and moments of LCT velocity have been recorded.: 

Each feature’s location, flux, size (pixels), and moments of LCT velocity have been recorded.

For a given AR, the velocity moments show a slightly negative divergence, but no net (signed) z-comp. of curl. : 

For a given AR, the velocity moments show a slightly negative divergence, but no net (signed) z-comp. of curl.

Aside: What is flux cancellation?: 

Aside: What is flux cancellation? Livi, Wang, & Martin (1985), in studies of photospheric magnetograms, defined it phenomenologically: “mutual apparent loss of magnetic flux in closely-spaced features of opposite polarity” 1 kword, from Bellot-Rubio & Beck, 2005: Martin (1998) observed that cancellation of magnetic flux at the photosphere is “necessary” for prominence formation.

Automated characterization of cancellation between features is especially interesting, but hard!: 

Automated characterization of cancellation between features is especially interesting, but hard! Emergence & spatial dispersal prevent reliable estimation of cancellation rates from summing over the field of view. Hence, FT is used. Possible criteria to discriminate cancellation: proximity, e.g., +/- features within a few pix both decrease in flux across time step; approach each other across time step.

Slide27: 

But these criteria are probably overly restricitive: fluxes fluctuate in time, and elements’ positions exhibit jitter. Does anyone have ideas to cope with these difficulties?

Further, projection effects in our LOS data can introduce spurious features!: 

Further, projection effects in our LOS data can introduce spurious features! Vector magnetic field data from Hinode/FPP and SDO/HMI should ameliorate this problem.

Each element’s displacement, dx, can be decomposed into components parallel & perpendicular r, the vector to the opposing center-of-flux. : 

Each element’s displacement, dx, can be decomposed into components parallel & perpendicular r, the vector to the opposing center-of-flux. This approach provides a statistical measure of shearing (right-hand sign convention) and convergence / divergence. dx r

The center-of-flux location is stable for long periods, as over 3.5 days in AR 7981.: 

The center-of-flux location is stable for long periods, as over 3.5 days in AR 7981.

Non-zero average values for shear & convergence can be seen for some ARs (units: Mx · cm/sec). : 

Non-zero average values for shear & convergence can be seen for some ARs (units: Mx · cm/sec). Avg. values provide estimates for how long it would take to, say, shear 1022 Mx by 25 Mm – about 106 seconds (~ 10 days) for AR 7981.

Conclusions are few! Our analysis is too preliminary to make any statistical statements. : 

Conclusions are few! Our analysis is too preliminary to make any statistical statements. But you can help us! Got any bright ideas? Are there other statistical measures that could be relevant to prominence/ CMEs? e.g., rotation of elements about the same sign center of flux – roughly, twisting motion size & flux distributions vs. AR disk appearance If you want to coordinate research on a given event, please get in touch.

Cartoon-ologically, cancellation is: 1) ideal submergence, or 2) ideal emergence, or 3) “reconnective.” : 

Cartoon-ologically, cancellation is: 1) ideal submergence, or 2) ideal emergence, or 3) “reconnective.” t1 t2 t3 1) 3) 2)