So, what is an NRAO Data Analyst? ... Well, the answer is non-trivial so I thought I'd put together a page that might be somewhat informative.
For people who just wanted to go to the Analysts' Webpage, but ended up here for some reason click HERE.
In a real sense, we are scientific helpers. We have many specific jobs that help in the day to day operation of the Very Large Array and Very Large Baseline Array. Much of what we do doesn't have much to do with data analysis. Data management may be an equally applicable term to what a Data Analyst does. We help move the flow of information and data along so that the two radio telescopes run on a fairly smooth schedule, and we aid in sending researchers observed data.
The specific tasks of a Data Analyst won't make a whole lot of sense until you understand a little about how the Very Large Array and the Very Long Baseline Array work. So, here is a general guide about each of the telescopes:
An astronomer would like to observe an object in space at radio wavelengths because this information will help in understanding how some object in space works, or what it is made of. The astronomer would like to use the VLA and sends a proposal of what he or she would like to observe to a committee that decides if the observation is worthwhile. If the astronomer's proposal is accepted, he or she must send an observing file to the VLA operators which tells the VLA what to look at and how to look at it.
The VLA is a connected network of twenty-seven, 82-foot parabolic dish antennas which are arranged in a 'Y' pattern. The VLA is called an 'interferometer' which describes how it processes the radio light from space so that the effective size of the radio telescope is that of the entire array. Each dish antenna receives a radio signal from some 'source' in space. A 'source' is a general term used to describe anything that is emitting radio light. A source could be a galaxy, quasar, pulsar, maser, star, etc. The radio light is collected individually by each dish and is sent to the main control building where the light from all the dishes are 'correlated' together. The data produced from correlation is written to magnetic tape and archived.
Once the correlator data is recorded to magnetic tape, the astronomer can ask to have a copy sent to him or her. When the astronomer gets the data, he or she will use various computer programs which takes the data and turns it into an image of what was observed or some other form of useful data. If a significant conclusion can be draw from the data about the object observed, then the astronomer may publish a paper describing the observation, the results, and the conclusions made.
The VLBA operates on the same principle as the VLA with a few addition complications. The VLBA is an interferometer with ten 'stations' (i.e. antenna dishes) scattered across the United States from the Virgin Islands to Hawaii. Thus, the VLBA works as a radio telescope as big as the US. However, since each antenna is so far away from each other, their signals cannot be correlated during a observation. Each antenna's data is recorded on magnetic tape during an observation and sent to the Array Operations Center where the tapes from all the stations are played back together and correlated. The correlated data is recorded on a small magnetic tape and sent to the astronomer who was granted the observing time.
A Data Analyst is a scientific associate specializing in tasks for VLA and VLBA operations. Data Analysts are seen as informal advocates for the researcher using the two arrays. They are charged with generating correlator scripts for VLBA projects as well as managing the SCHED control files needed for VLBA observations. The correlated data is initially examined by the Data Analysts before it is sent to a scrutinizer for more detailed examination. Data Analysts also check observe files for the VLA and run baseline and pointing tests to help optimize the performance of the instrument.
Buzz words will be italicized.
April 15, 1998.
I arrived at work around 7:45 in the morning and finished copying a VLA data set that I had started the day before. A print-tape had to be run on the DAT tape that had the data on it, and a shipping form had to be completed. The data was requested by a researcher in Germany working at the Max-Planck Institute. Each VLA observing project is denoted by a two-letter, three-digit proposal code beginning with an 'A', such as AW984.
Next, I started on what I call 'morning chores'. I checked the analysts' Email account to see if there were any emergency Emails that needed to be dealt with immediately. Then I went to the mail shelves to check if there were tapes from the VLBA correlator with data ready to be sniffed. There were four projects: TR007, TC005, BL064, and GG034. Regular VLBA projects are designated with a 'B', VLBA tests begin with a 'T', and projects including radio telescopes from Europe or other parts of the world are designated with a 'G' for Global. I brought the tapes back to the office and recorded their arrival in our queue and wrote them up on the dry-erase board so we'd all know what was ready to be sniffed. I checked and read my own Email.
Noticing a VLA data request on another white board we have up in the office, I grabbed the VLA logs, went to the tape archive vault, and checked out Exabyte tapes for two projects called AB865 and AH603. Using a program called VLAR on my computer, I copied the data to another Exabyte tape.
I took the VLBA DAT tape for project BL064 and placed it in the DAT drive on my workstation. I set things up to sniff the data on the tape and started filling out a checkout list for the data. Alex, a fellow data analyst, starting sniffing TR007 and TC005. Meri Stanley, the head analyst, took the redistibution of GG034 to sniff. Setting the sniffing program in motion, I went to check to see if some old 9-track tapes from the VLA had arrived. They were needed for a VLA data request that I started a few days ago, but couldn't complete since the 9-tracks had to be ordered from the VLA site itself. They weren't in yet.
I went back to the office, and the head correlator operator, Steve Thompson, was there discussing the current correlator projects with my supervisor, Meri Stanley. Meri listed the VLBA projects that were ready to have job scripts made on the white board. This meeting is know as the 'calf' meeting since it is a baby 'COW' meeting. The 'COW' meeting is a weekly meeting where the status of current observed VLBA projects are reviewed.
I began generating jobs for a VLBA project called BD046W. It is a spectral-line project monitoring Silicon Monoxide masers. This particular observation included only the standard VLBA stations. Usually, it also includes a single dish VLA antennae but not this time. I filled out the checklist listing all the scripts that needed to be run and some places in the files to check. This time there were no big problems that required extra attention.
By that time, the Exabyte data copying has finished, so I ran a print-tape task, and prepared it for shipment.
The sniffer program finished reading the data for the VLBA project BL064, so I looked over the plots that it generated and noted anything that may have corrupted the data on a check-out sheet. I then passed the DAT tape and the project file on to Ketan Desai, a staff scientist who was the scrutinizer for this project.
I went home for lunch.
After lunch, the 9-track tapes had arrived from the VLA site. So, I first cleaned the 9-tracks with two tape cleaning machines in the main computer room. This took eleven years worth of oxidation and gunk off the tapes. Next, I set up VLAR on a workstation in the computer room that had both a 9-track reader and a DAT tape drive.
After starting the first 9-track copying, I headed back to my usual workstation to start a big Exabyte copy request, after gathering all the archive tapes needed from the vault. Then, I basically spent the next few hours running back and forth, between the two places that I was copying tapes, to change tapes when the previous ones were finished.
Since Meri works part-time and leave at 2:30 in the afternoon, Alex and I kept vigil over the Analyst and Vlbiobs Email accounts in case an urgent Email should pop up.
Here is a list of some characteristics that will help a data analyst succeed with his/her work.
Last modified: July 11, 2007
Last updated: April 28, 1998
Written by Jason Wurnig, revisons by Lisa Foley.