1 / 14

Bodo Reinisch 1 , Ivan Galkin 1 , Shing Fung 2 , Robert Benson 2 ,

Towards the VWO Annotation Service:. A Success Story of IMAGE RPI Expert Rating System. AGU FALL MEETING • San Francisco, California, USA • December 15, 2010. Bodo Reinisch 1 , Ivan Galkin 1 , Shing Fung 2 , Robert Benson 2 , Alexander Kozlov 1 , Grigori Khmyrov 1 , and Leonard Garcia 2

aniles
Download Presentation

Bodo Reinisch 1 , Ivan Galkin 1 , Shing Fung 2 , Robert Benson 2 ,

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. Towards the VWO Annotation Service: A Success Story of IMAGE RPI Expert Rating System AGU FALL MEETING • San Francisco, California, USA • December 15, 2010 Bodo Reinisch1, Ivan Galkin1, Shing Fung2, Robert Benson2, Alexander Kozlov1, Grigori Khmyrov1, and Leonard Garcia2 1University of Massachusetts Lowell • Department of Environmental, Earth, and Atmospheric Sciences • Center for Atmospheric Research 2NASA Goddard Space Flight Center Acknowledge: RPI BinBrowser software

  2. Challenges of Wave Data Interpretation“Wave data don’t come with names” fce fpe fuh Wh AKR Z 1Sx 1Sozo fx fz FUV on IMAGE showing brightening of aurora emissions Hinode telescope showing multiple solar flares RPI plasmagram taken in 60 -1000 kHz showing… hmm… what is this?

  3. RPI Plasmagrams of Special Interest Contain signatures of active sensing remote plasma regions Recorded at opportune times and locations on orbit

  4. RPI Data Avalanche • 1.2 million plasmagrams acquired • At 5 sec/image, working 20 hr/week, data exploration takes 2 years • Realistically, one summer student can process and tag 10,000 images • 30 student-years are needed DATA EXPLORATION IS EXPENSIVE Discovered data need to be organized and preserved

  5. BinBrowser Main Screen A B D C E (A): Image, (B): Ratings, (C) Annotation Selector, (D) Comments, (E) Expert view

  6. BinBrowser Query Dialog (F): Time Selector (G): Expert selector (H): Query by rating (I): Query by image attribute Expert Rating System enables queries by phenomenon 6

  7. Results and Lessons Learned • RPI Ratings System holds 7,530 expert annotations • Strong educational benefit for users • Exemplary plasmagrams with descriptions provided by domain experts • Strong organizational benefit for experts • Easy retrieval of plasmagrams by content (event or phenomenon type) or by context (spacecraft location, geospace conditions). Examples: • Find all plasmagrams with epsilon signatures • Find all plasmagrams with traces inverted to density profiles • Find *my* rated plasmagrams recorded within the plasmasphere • Data Discovery benefit – yet to be discovered in full

  8. Data Discovery via Rating System Data Discovery using automatic image prospecting • Neural network model of pre-attentive human vision • Selected 225,124 plasmagrams with echo traces • Success Stories of Data Discovery • Find plasmagrams with echoes propagating in the whistler mode • Find sequences of plasmagrams with traces within one plasmaspheric fly-through (3,365 orbits found) • Find plasmagrams with echoes within the trough • Find plasmagrams with large number of echo traces CORPRAL Cognitive Online Rpi Plasmagram Rating ALgorithm

  9. Lessons Learned • Students have embraced new technology • Senior scientists encouraged to use new technology • Notable ERS followers do exist • Technical problems shall be addressed to help • Database access restrictions (firewalls) • Latency • We need to place rated plasmagrams into the context of other space physics data resources (outreach)

  10. Lessons Learned (2) • Psychology of annotating data • ANNOTATE AS YOU ANALYZE • Annotation interface has to be part of data analysis environment to allow ergonomic solutions for an “on the spot” annotation of data • “Incubation” of annotations? to prevent premature release of research ideas • Implemented in BinBrowser as “public or private” tag • Give annotations status of the intellectual property (keep track of the authorship and submission dates) Would you annotate a significant data discovery in a public VxO (before it is published)?

  11. AKR AKR Solar Type III Radio Burst KC Upper Hybrid Resonance Magnetosheath Noise Continuum Radiation PH PH (n+½) Gyroharmonics

  12. Wave Phenomenon Ontology: Preview Auroral Kilometric Radiation (common name) • Class = Radio Emission • Qualifier = Natural • Propagation = Free Space • Propagation Mode = R-X • Frequency Character = Wideband • Spatio-Temporal Character = Continuous • Origin = Earth.NearSurface.AuroralRegion • Wave Type = Electromagnetic • Spectral Band = Kilometric • Observed Region = Earth.Magnetosphere

  13. Summary • An Expert Rating System was developed for 1.2 million dataset of IMAGE RPI plasmagrams to annotate observed phenomena • 7,530 plasmagrams have been rated by human experts • 225,124 plasmagrams with specific features were identified by CORPRAL data prospector • Expert Rating System has educational and organizational benefits • Data Discovery capabilities are being explored • Rating System to be considered as a prototype for VWO Annotation Service

  14. Data Exploration and Analysis Expert Analysis Scaled traces, matched resonances Inverted profile of plasma distribution Color scale for optimal X antenna display Default visualization

More Related