Astrofriend's homepage
Share: Twitter Reddit Facebook Gmail Gmail Email
Search Astrofriend's homepage:

Valid CSS!

All pages shall now have been validated

Info Cookies (Kakor) / GDPR


Advertisement /

Twitter @AstrofriendLars

Follow Astrofriend

Sample of Project News

Advertisement / Annons:

Astronomy Science data


  1. Overview

After had done a couple of test with data from the POSS archives data I wanted to do something more complicated and with modern digital sensors. After I had investigated the different archives I found the Kepler archive was the most interesting to do the next test with.

As you maybe already know this telescope is specialized to search for exoplanets. With it's 10 by 10 degrees wide field camera it looking at about 100'000 stars that they suspect to have exoplanets. They take a photo (or collecting data as astronomers say) every 30 minutes.

Not very interesting for me to investigate the exoplanets data, that's what all other do. Can I use the data for something else? I have a weak memory of that I many years ago read something about that these data could be excellent to haunt for variables too, that's stars that are not stable, they have a variation in the light output. That sound much more interesting for me to investigate. But how to do it?

After reading about how the Kepler manage the data I see that it doesn't send full frames of the field down, it's just pickup a very small area around the stars it investigate, something like a 20 x 20 pixel area. That is something I can't use. But there is one another alternative. Kepler sent down the data once a month and when sending all that data down it's also sent one complete field from all its 2x42 sensors, a big chunk of 400 Mbyte that contain all the 84 sensors data. Great, that something I can use, not as good as 30 minutes time separation, but monthly data. After all, it's a space telescope, no clouds!

Kepler Spacecraft:

One big advantage with this telescope are of course it's up in the space. No clouds and collect data almost 24 hours per day. When reading about the telescope I see that it's very similar to the telescope that take the POSS data. A Schmitd telescope with 0.95 m diameter (38") opening. You can read more about it here:

Camera sensor:

The sensors cover all together 115 square degrees or about 10 x 10 degrees. There are 84 sensor areas from 42 sensors of 2k x 1k sensors, all together they produce about 94 mega pixel data each exposure. Each pixel is stored as 32-bit floating point data.

Because of the very big total size of the sensor this telescope behave as relative wide angle lens, at least when you talk to an astronomer. The fov (field of view) or angle that is captured by my 150 mm lens on my full frame camera is almost the same. But my camera doesn't have 90 million pixels and then not have that resolution, and a small diameter lens doesn't give that resolution either. The Kepler camera is like a DSLR camera on steroids.

These data was much more complicated to handle, all data are stored in a huge 400 Mbyte fits file and I had to select out each sensor's data from that. It's even more complicated because the satellite rotates 90 degree about four times a year. You have to search for your object then from four different sensor areas, then rotate and mirror and add extra area to get them in equal size.

More to read:

These CCD sensors doesn't look to have overflow channels, as you see in the image I have made from this data stars are easily over exposed with only 6 seconds exposures (the sub image of the 27 minutes exposure). An overflow channel would have make the sensors less effective because it take up space where you can have the sensitive area to collect photons with. With a low noise readout circuit you can do many readouts and still have low noise and gain dynamic.

Short list of what can be good to know about the Kepler instrument.

  • Schmidt telescope: 0.95 m opening, 1.4 m focal length, f/1.5
  • Field flatter lens, one above each sensor?
  • Each sensor have 2200 x 2048 pixels
  • Pixel size; 27 x 27 micro meter
  • Well depth: 1x106 e-
  • QE: >0.8
  • Total number of pixel 94.6 million divided at 42 sensors, divided in two, total number of areas 84
  • Gain: 110 e- / DN
  • Max signal: about 12000 DN, 14-bit resolution?
  • Pixel scale: 3.98 arcsec / pixel
  • Each 27 minutes of data consist of 270 numbers of 6.02 seconds exposures
  • Cumulative readout noise: 81 e-
  • Max stacked ADU count before overexposure: About 200'000
  • Stored unit: electrons / seconds
  • Stored data resolution: 32-bit floating point FITS data
  • Wave length: 420 to 900 nm
  • Timestamp at middle of exposure
  • Time format: Modified Julian Date corrected for gravitation center of solar system

What can be done with the Kepler data archive:

I took me many days to figure out how to get the data in a form that I can do image processing on. Here are the first data out, just a single image from a single sensor area:

Next goal was to align and stack the 44 sub images:

As a first step in the variables world I have used these data to make a movie ranging over the almost four years that the Kepler space craft was collecting it's data.

Now when understanding how to analyze variables a bit I search of other things that could be interesting to analyze with the Kepler data. One of Kepler spacecraft specialties is the high precision of the measurements it do. It's designed to find the tiny variations when exoplanets obscure a part of the starlight. Here I try to find stars that behave different relative it's neighbor stars.

That was not bad, the third object I investigated was an unknown variable. It was only possible because of great help from Hans and Thomas to find it was not discovered yet.

I found this work with variables stars really interesting, but have no experience of it. Easy for me to find stars that show up patterns that place them beside the normal stars. But then I found it at this stage not so easy to identify the star and find if it's an unknown variable. I need help, I asked Hans Bengtsson and Thomas Karlsson if they want too cooperate about this, and they did.

Now with help from Hans and Thomas I relative easily can identify the stars. What's left is to verify if it's already discovered as a variable or not, and if not to register the discovery. Hans and Thomas help me with that now in the beginning.

Variable amateur astronomers normally use the JD (Julian date) as timestamp of the data, professional astronomer use the MJD (Modified Julian Date). If you want to compare the data it's wise to use the same standard.

Because I have both rotate and mirror the images the orientation arrows up in the left corner is misleading. The north direction is about in the 11 o' clock direction.

More will come later, I still missing knowledge about some parts.

Go Back to content

Go Back

Advertisement / Annons: