1 / 11

Questions for Technique Sessions Chair(s): Hill, Schou, Larsen Status: [draft]

Questions for Technique Sessions Chair(s): Hill, Schou, Larsen Status: [draft]. H6: Global Helioseismology Techniques & Data Products. Thompson – inversion Main task: making the pipeline robust, automate Subtle feature, how to distinguish noise realization from real signals?

twayne
Download Presentation

Questions for Technique Sessions Chair(s): Hill, Schou, Larsen Status: [draft]

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Questions for Technique Sessions Chair(s): Hill, Schou, Larsen Status: [draft] H6: Global Helioseismology Techniques & Data Products

  2. Thompson – inversion • Main task: making the pipeline robust, automate • Subtle feature, how to distinguish noise realization from real signals? • take serious what are the confidence levels? • Beat systematics, estimate error magnitude & correlation • Near surface • High-l data • Incorporate local inversions? 3D models • Test inferences with data from C.Z. simulations • Korzennick: Need to re-evaluate peak-bagging (has stagnated) • Need multiple methods that incorporate what we learned from • Larsen: Incorporating a-coeff. correlations makes little difference in inversion and error bar.

  3. Rachel Howe: Peakfinding & inversions • GONG peakfind code. “Rusted in place” very hard to convert. IRAF • Need for HMI • Leakage matrix • Standard products • Standardize mesh, trade-off (on demand with custom parameters) • Post launch: • Cross-check between • Basu: Inversion parameter choice is a problem. Need scaling with depth. • SOLA (an RLS) vulnerable to bad/sparse data points.

  4. Ed Rhodes: High degree fitting. • Many improvements in ridge fitting technique. • Main holdup for integrating into pipeline: Need to manually check input “seed” values. • Discussion: • Sylvain: Asymmetric fits introduce correlated parameters in fit. • Mark Rast: Fitting heuristic asymmetric profile is useless without a consistent model that explains the entire l-nu diagram from proper understanding of source parameters.

  5. Sylvain: Low degree fitting code • Introduce instrumental one at a time to study effects. Gives confidence that error estimates are reliable (the model predicts the right systematics). • V & I comparison • Erro budget: • Confidence in correction must be higher than uncertainty of fit. • Hurdles • Fix the obvious: plate scale, leakage matrix • Need to know: tilt, plate scale, PSF • Knowing P0, B0 better • No. 1 priority: pre-launch instrument tests. • Post-launch validation • Discussion: Rast: I & V crucial.

  6. Stuart Jeffreys: peak bagging • Leakage matrix & spectral line profile • Keep putting effects into leakage matrix - when do you stop? • How high I L without ridge fitting? Up to ~250-300 • Interpretation of parameters, e.g. re-parameterizaton to reduce number of parameters. How to interpret. • Model all spectrum? • Use high S/N data to improve algorithms • Reduce # of params (100 parameters for the whole spectrum) • Ultimate goal: • #1 goal: refine leakage matrix.

  7. What do we need to make progress on the science questions in general? Should a TR&T focused topic be made? Probably not, too technical, unless outside of LWS SR&T instead? Working groups Probably useful for fitting Inversions with high degree modes Surface treatments, kernel calculation Schedules, meetings, splinters, etc. +1 month: Status review, plan of attack, priority setting +6 months: Progress review, selection of HMI modules +18 months: CDR of selected modules +24 months: Acceptance testing Observables, models, codes, resources, people … Observables: V+I time series, phase differences, coherence Models: Near-surface (first 1000 km?), excitation, spectral line profiles Codes: Leaks, ridge fitting, V+I, “Global Global” fitting, direct inversion of spectra Artificial data (gasp!) Resources: CPUs, disk, $$, time People: Hill, Howe, Schou, Rhodes, Korzennik, Jefferies, Burtseva, Salabert, Thompson, JCD, Vorontsov, Tripathy GH Implementation: general

  8. Identify hurdles, bottlenecks, uncertainties: Observables: No HMI proxy data, but plenty of MDI & GONG data Models: Need near surface treatment Codes: High-degree ridge fitting is very expensive, can we invert high degree data? Resources: will need large amounts of compute time Artificial data is a quagmire People: Everyone’s busy! Identify best codes and approaches and/or selection criteria High-degree ridge fitting Leakage matrix calculations V+I fitting “Global Global” fitting Main technique hurdles: High computational costs Verification of results Simulation artifacts Identify standard products and specifications that are needed as input to the technique code (report to D sessions) Observed parameters Simulated parameters Identify external (to SDO) data products that should be available GONG products MDI products Simulation products GH Technique Development

  9. Data Product Development Session Chair(s): Hill, Schou, Larsen Status: [draft] H6: Global Helioseismology Techniques & Data Products

  10. What data products must we produce? Frequencies, amplitudes, widths, excitation parameters, errors for 0 ≤ ℓ ≤ 2000, and 0 ≤ ≤ 10 mHz. All ms or a-coefficients or ? Inversions of rotation rate, sound speed differences for 0 ≤ r/R≤ 1.1 Leakage matrices Inversion kernels Assessment of required resources, etc: Pipeline software Two major independent fitting codes available, both need work Ridge fitting & global global are major research tasks Analysis software/studies Inversions & kernels probably well in hand now But surface effects remain Supporting software/models Leakage matrices are a major task Computational requirements (run time estimates, system requirements, …) Very large computational tasks, get as much cpu and memory resources as possible Storage requirements: size, duration, etc. Data set size is small, duration should be indefinite Access: web, archive, logs, search methods, etc. Make sure processing details are archived, provenance is important here. HMI GH data products

  11. What codes do we have now and what codes need to be developed? Have MDI & GONG fitting codes routinely running, both could use improvements Need ridge fitting and 3-D global global codes Have MDI & GONG leakage matrix codes, both could use improvements Have kernels & RLS, OLA inversion codes for interior Need inversion technology for near surface regions Who will can implement the required codes? Ridge Fitting: Rhodes, Reiter, Korzennik, Rabello-Soares Global Global : Jefferies, Vorontsov Leakage matrices: Hill, Schou, Burtseva, Howe Current fitting improvements: Hill, Schou, Burtseva, Howe, Salabert Inversions: Thompson, Howe, Corbard, Schou, JCD Surface Kernels: Thompson, JCD, Jefferies, Vorontsov Define key milestones, test procedures, and target dates, … + 1 year: Current fitting & leakage code improvements completed. Tested with simulated time series +2 years: Ridge fitting and Global Global codes completed. Tested with simulated power spectra +2 years: Near surface effects included in models and inversions. Tested with Hare & hounds exercise Communication: define or list meetings, topical sessions, etc., where progress can be presented, discussed, evaluated, … SPD New Hampshire, June 06 SOHO/GONG/HELAS, Sheffield UK, Aug 06 SPD, Hawaii, summer 07 Set up a “GloHCo” group, with meetings every 3 months until launch GH Implementation X

More Related