Cosmic Variability and the IT Challenges Mass Data Management of - PowerPoint PPT Presentation

Cosmic Variability and the IT Challenges ─ Mass Data Management of Astronomical Databases Wen-Ping Chen Institute of Astronomy National Central University Taiwan 2006 May 04 @ ISGC2006 1

Outline Outline Fully operational; in Taiwan TAOS Taiwan-America Occultation Survey Pan-STARRS Panoramic Survey Telescope and Rapid Response Being constructed; System in Hawaii, USA 2

3

L ULIN O BSERVATORY 鹿林天文台 Elevated to 2862m; above inversion layer … seen from Yusan (Jade Mt) 4

LELIS SLT LOT TAOS LULIN OBSERVATORY 5

The TAOS (Taiwan-America Occultation Survey) project, a novel telescope array set up by groups from Taiwan, US and Korea, began Comet nuclei too faint to routine observations in early 2005 be detected by direct and has the potential to make unique imaging may be “seen” when they move in front of contribution to the knowledge of our a background star --- a stellar occultation event. Solar System. 中美掩星計畫 6

Project Overview � Census of the solar-system family � An array of wide-field telescopes (D=50 cm, f/1.9, FOV=3 sq. deg) to monitor the brightness changes of ~1,000 stars at 5 Hz rate � Looking for a ‘blink’ of starlight (occultation) when an object (> 2 km) moves in front of a distant star Frequency of events � population of “interveners” � Data rate a few 100 GB per night; only “interesting” data downloaded via the dedicated E1 connection � Real time data analysis (light curves, statistics) � Requiring coincidence detection of the same event by all the telescopes (against false positive) 7

TAOS Telescopes TAOS is the only one of its kind in the world to Lulin Observatory conduct a census of small (1-2 km size) icy altitude=2862 m bodies at the outer reach of the solar system. With a special data acquisition and a non- parametric statistical 100 GB/night analysis scheme C B D A 8

Adaptive Aperture Photometry Pipeline � Fast � Moderately accurate � Compensate for image motion 9

Sample Output of the TAOS Photometric Light Curves beg. time end time count ct err x y rb 10

• Pipeline Flow from image taking to archival • Arrows represent the flow of messages between components STATISTICS STATISTICS CAMERA CAMERA CAMERA CAMERA CAMERA CAMERA CAMERA CAMERA AGGREGATOR AGGREGATOR PHOTOMETRY PHOTOMETRY FITS FITS CARTOUCHE FITS FITS FITS FITS FITS FITS ARCHIVER ARCHIVER SCHEDULER SCHEDULER DB DB 11

T EST D RIVE 1 CY Aqr, a known Delta-Scuti star with P~88 min, was observed by TAOS on 2003 September 16 with 0.3 s sampling, here binned to 150 s for illustration. � time-domain astrophysics 12

T EST D RIVE 2 2004 February 21 TAOS detected the occultation event of HIP 079407, m V =8.8 mag) by Δ t ~ 6.25 +/- 0.50 s (51) Nemausa (m V =11.9) Prediction by Isao Sato ( 左藤勳 ) D~150 km 13

T EST D RIVE 3 2004 June 05 TAOS detected the occultation of HIP 050525 (m V ~8.46 mag) by (1723) Klemola (m V ~15.7 mag; D~31 km) with two telescopes TAOS/A TAOS/B Enclosure opened by a resident assistant and observations carried out remotely from Taipei 14

T EST D RIVE 4 2006 Feb 06 three TAOS telescopes detected the occultation of TYC 076200961 (m V ~ 11.83) by (286) Iclea (m V ~ 14.0 mag, D~ 97 km) 15

Data Acquisition Data Acquisition Typical CCD imaging Every star, together with surrounding skies, get exposure at the same time TAOS data Integrate for 200 ms and then read out 32 rows of pixels, with the shutter remains open The sequence continues, so each star appears as a series of dots � ‘zipper’ ‘Fake’ neighboring stars and skies! 16

E E vent Detection --- vent Detection --- Rank Statistics Rank Statistics Use the rank, instead of the flux, to quantify the light curve 4 Π = − 4 Z log ( S ) log ( W ) w 10 10 i = Simulated light curves from i 1 each of the four telescopes A true occultation event should have the lowest rank in all telescopes no need for highly accurate flux � speed With occultation Without conditional probability Rank statistics � low false rates 17

An event can be detected even it is not obvious in the data � Higher flux ranking 18

Panoramic Survey Telescope and Rapid Response System 19

Project Overview Wide-Field Imaging � All-sky survey (3 π ) Short Duty Cycle � Frequent revisit (cadence 4-7 days) Efficient Operations � An array of 4 telescopes, located in Hawaii, each of D=1.8 m, equipped with a 1.4 gigapixel camera of an Orthogonal Transfer Array CCD detector (=40 cm square focal plane) � 7 square-degree FOV with 0.26” pixels � Detection of moving, transient, and variable celestial objects down to very faint limits � Cumulate very deep sky images 20

The Sciences • First large-field survey program to open the time domain in astronomical observing ------ transient sources in time & space • Multiple survey modes, both wide field ------ ecliptic (& “ sweet spots ”), all sky (3 π ) ------ and selected deep fields: – Solar System (PHA emphasis) – Cosmology (weak gravitational lensing, supernovae, GRBs) – Galactic Structure • Ultra-deep static images (R < 23.5 mag) 21

Pa n-ST ARRS Mino r Pla ne t Summa ry 8 7 10,000,000 6 1,000,000 5 100,000 Series1 Known 4 Series2 10,000 PS 1 Ye a r Series3 3 PS 10 Ye a rs 1,000 2 100 1 10 0 1 NEO / PHO Jovia n T Othe r T Ce nta urs Come ts Ma in Be lt T Wide T Compa nions Inte rste lla r Visitors 1 2 3 4 5 6 7 NOs 8 9 10 NO Bina rie s roja ns roja ns 22

23 The Telescopes

The Detector Independently addressable orthogonal transfer CCDs (cells) � Reducing cost by increasing yield � Fast readout: Gigapixels in 2 s � On-Chip guiding � Minimizing effects of bright stars � Compensating for image motion 24

OTCCDs Work Note: the arrows point to two examples in the images where the improvements in image quality due to OT tracking are clearly evident M13 I band 300 sec With OT tracking Telescope guiding only 0.45" FWHM psf 0.59" FWHM psf 7 Hz frame rate 25

26 Prototype telescope site --- Haleakala High Altitude Observatory (Maui) The Site(s)

27

28 Eventual Mauna Kea site for Pan-STARRS

The Budget � $60M funded by Congress through US Air Force (2002) � Funding for construction only --- Telescopes, Detectors, and “working” control software and analysis pipelines � Annual operations cost amounts to ~$2M/year � Operations partners sought to share cost and make use of data for breakthrough science � In-kind contributions encouraged --- technical human power, e.g., professional software engineers, observers � Scholar and student exchanges 29

IT Challenges Each raw image from a single Pan-STARRS camera will contain 2 Gbytes (2 bytes per pixel). In the full survey mode, typical exposures last 30 seconds , so the raw data rate is several terabytes per night for the full telescope system. The amount of data produced by Pan-STARRS is so large that it will not be practical to archive every image. Software techniques are therefore being developed to extract the important information from the images, while allowing less crucial information to be discarded. 30

Pan-STARRS Pan-STARRS （ 泛星計畫） Panoramic Survey Telescope and Rapid Response System U of Hawaii 4 x 1.8 m telescopes Orthogonal Transfer Array CCD technology (G pix) � 7 sq deg Whole sky patrol every 4-7 days down to 24th mag several TB/night … a 10-year movie of the cosmos � moving, transient, and variable objects 31

The Data Flow Subsystems � TEL – Telescopes � CAM – Cameras � OTIS – Observatory, Telescope & Instrument Software � IPP - Image Processing Pipeline � MOPS – Moving Object Processing Software � PSPS – Published Science Data Products 32

Summit Process Flow Diagram ��������� ������ ������� "������� ���� ������� ���������� #������������� ���������� ��������� �������� ��������� ��� �������� ���� ����� $%������ ������� ���� ������� ������ ������� ����������� ����������� ������ ���� �� �� ������ ���� #������ �������� !��� ���� �������� ��������� ���������� ( ���������� �������� ���� ������� ��� ������ ������ &����� �� �� ������ ��%�� ���� ��������� !���'��� #������ �������� ���� �� $%������ �������� ���������� ������� ������ 33