Download presentation
Presentation is loading. Please wait.
Published byCrystal Casey Modified over 9 years ago
1
1 Brent Harris, 3 Anthony Remijan, 1 Kevin Lehmann, 1 Brooks Pate 1 University of Virginia; 2 Virginia Image and Video Analysis; 3 National Radio Astronomy Observatory Identifying Local Chemical Environments in Orion KL by Broadband Data Cube Analysis
2
Orion KL EVLA Demonstration Science K band WIDAR: 3 x 1GHz, 1.5hr integrations, 12 Dishes (Dec, 2009) Interacting with Image Space ~10 kPixel Pixels: 96 X 96 (1’ x 1’) Synthesized beam: ~1.5” Interacting with Spectral Space Data Channels (Frequency): 24,012 1.5km/s or 125kHz resolution Bandwidth: 23.6 – 26.6 GHz Final image format illustration 10,000 Spectra 24,012 Images File Size: 885 MB (~220 million data points) Spectrum: Everything behind a pixel
3
Broadband Spectral Diversity Chemical variation Energy variations Absorption/Emission
4
Orion KL ALMA Demonstration Science Band6: 5 x 8GHz, 15min integration, 22 Dishes (Mar, 2012) Interacting with Image Space 10 kPixel Pixels: 100 X 100 (40” x 40”) Synthesized beam: ~0.6” Interacting with Spectral Space Data Channels (Frequency): 76,800 0.6 km/s or 500kHz resolution Bandwidth: 213.7 – 246.6 GHz 10,000 Spectra 76,800 Images File Size: 3 GB (7.680 billion data points) Raw data > 30GB What is behind the Hot Core pixel???
5
Broadband Spectral Diversity 2 data sets and already a challenge
6
Fundamental Science 1) What chemicals are present in the ISM? (1D question) How to use all dimensions at once??? 2) What are the distinct chemical environments? (2D question, spatial resolution) 3) What is the unique composition of the chemical environments? - parent molecules and the ensuing chemistry - A 3D BROADBAND PROBLEM, BUT USUALLY TURN IT BACK INTO A 1D PROBLEM Using the broadband interferometers
7
1 What Chemicals are Present? Spatial information increases confidence in line assignments… SO 2 8 17 – 7 26, E L = 35K ??? SO 2 8 26 – 9 19, E L = 42K Want to QUANTIFY the correlation 8 26 – 9 19 Distribution Flux (Jy/bm) 45 0 0 0 96 Pixel 8 17 – 7 26 Distribution Flux (Jy/bm) 45 0 0 0 96 Pixel
8
Sum Methyl Formate Channels Flux (mJy/bm) 130 0 0 0 96 Pixel Sum Ammonia Channels Flux (Jy/bm) 45 0 0 0 96 Pixel Sum OCS Channels Flux (mJy/bm) 140 0 0 0 96 Pixel Sum Ammonia Channels Flux (Jy/bm) 45 0 0 0 96 Pixel “Double Resonance” in Astronomy Image Correlation: (Pearson’s correlation coefficient) CH 3 OCHO OCS NH 3 r ≈ 1 r << 1 Take a cue from Digital Image Correlation applications… statistical analysis, security, document verification
9
Chemical Correlation r ≈ 1 Practically perfect correlation by Pearson’s coefficient Two oxygen containing molecules with similar distributions
10
Correlation Space Imaging the spatial and spectral information together -generate a coefficient for each pair of data channels (24012 2 ) Red – high correlation between two data channels Diagonal – each data channel perfectly correlated to itself Box width – line width of a spectral feature Featureless channels can be dropped
11
2 Distinct Chemical Environments *Axes are a non-linear progression of data channels (compression) Data compression to visualize the whole dataset (only bright data channels are retained) Correlation is NOT high between all spectral lines Indicates two different environments (images of these data channels are very different) NOT high between all spectral lines
12
Comparisons of spectral features *Axes are a non-linear progression of data channels (compression) First 8,000 data channelsFull data set compressed Two different emission features
13
Comparison of Spectral Features Ammonia Consistent Self Correlation Methanol Incomplete Self Correlation What are the images at line center?
14
Methanol Masers CH 3 OH Methanol Distinctly different image and narrow line width
15
Hot Core NH 3 vs Methanol Masers NH 3 Ammonia CH 3 OH Methanol Distinctly different image and line shape
16
Correlation of Spectral Wings What are the images in the wings?
17
Channel by Channel Imaging 80% correlation to “hot core” image (r=0.80) Sum these channels for each of 8 methanol masing lines. Can find many distinct environments in one spectral line. (B)
18
LTE Methanol LTE Methanol in Orion KL “Hot Core” Found the “hot core” methanol measured by Wang et al. 2011, A&A, 527, A95 (HEXOS) Thermalized at about 120K. Black: Methanol HC flux values Red: Simulated methanol LTE spectrum at 120K Separated Masing spectrum from LTE
19
3 Chemical Composition Co-spatial Chemistry All of these features are co-spatial Really a frequency axis
20
Extracting Co-spatial Spectra All emission that is correlated to the hot core (blue) can be extracted from the sum spectrum over all space (black). Extended emission does not contaminate. The doublets in the black spectrum (methyl formate) do not appear in the hot core.
21
Conclusions – a Data Enabled Age The computer storage and processing capabilities are barely compatible with data throughput at JVLA and ALMA. Working in “correlation space” is an innovative way to address the capabilities of new technology and answer fundamental science questions. -Gives the human eye a more comprehensive view of the data -Works in a numeric space that is compatible with automation -Reduces the data set to manageable sizes -Incorporates a statistical result opposed to qualitative analysis -Pattern matching in flux space and correlation space Correlation space analysis can be used for: -Aiding spectral assignment (pattern matching in flux or correlation) -Extracting co-spatial features -Identifying unique physical environments for a species
22
Acknowledgements Clare Yang Virginia Imaging and Video Analysis Crystal Brogan National Radio Astronomy Observatory Pate Lab Eric Herbst University of Virginia This material is based upon work supported by the National Science Foundation Graduate Research Fellowship under Grant No. DGE-0809128 Peter Schilke Karl Menten
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.