Kepler Channel Noise Problem

Sorry, I could sound naive, but is it possible to rename and shift this discussion in Detection Methods and Projects.

I mean, this thread has taken 10 pages for discussion about Kepler results, but so far I've heard nothing Kepler has detected or discovered.

Yeah, fair enough. Give me a sec while I round up the posts.

Of course getting enough time on HST for doing that kind of detection is another matter

Putting my moderator hat on here, remember if you are contacting the Kepler team to make sure you have permission to quote them before posting quotes from private emails.

Sirius_Alpha wrote:

Yeah, fair enough. Give me a sec while I round up the posts.

Indeed Corot results has taken a similar number of thread pages but discusses about detections.

So far Kepler seems having collected data and working not so good too.

Thank you for the attention

Also HST field of view much smaller

Edasich wrote:

Thank you for the attention

Any time =)

Edit:

Lazarus wrote:

Putting my moderator hat on here, remember if you are contacting the
Kepler team to make sure you have permission to quote them before
posting quotes from private emails.

I second that.

If not hard to ask a question William Borucki directly in the comments?

http://blogs.discovery.com/space_disco/2009/10/keplers-exoplanet-hunt-on-hold-until-2011.html
I can not, bad English.

Borislav, at least two of the people on this forum have already attempted contact with the Kepler team. I am sure we will hear something soon.

I have received email back from the Kepler team, I'm going to request quoting permissions, but the short story is that the doom-and-gloom scenarios are incorrect. Things are not nearly as bad as the news articles and posts upthread are making out.

More on this to come later...

Thanks, really good to hear and I look forward to hearing more.

Good to hear.

We are really curious about the rocky planet scenario.

We all agree the Kepler mission will produce ~ 1000+ planets.

It is the 50 or so rocky ones that worry us.

William Borucki, the principal investigator for the Kepler mission who asserts that the news article was 'incorrect' really needs to be more specific concerning the precise impact of the known bad amplifiers on the ongoing dataset acquisition pre-software fix. This issue should be covered in detail in the iminent Mission Manager's report. If we can get chapter and verse posted about microprocessor glitches, this far more threatening (or is it? we don't know the facts) issue needs to be explained. The fact that it has been a known problem for a year certainly should have given the team time to at the very least assess the impact.

http://www.newscientist.com/article/dn18095-telescope-glitch-could-delay-discovery-of-alien-earths.html?DCMP=OTC-rss&nsref=online-news

More Borucki comments.

Still skirting the issue however. Wish he would take it head on and address it. Esp. with respect to the differential photometry.

Quote :

Three of these channels are plagued by electronic noise that makes
stars in their field of view appear to flicker – "like it's changing
its brightness at a rapid rate", says Kepler chief scientist William
Borucki of NASA's Ames Research Center in Moffett Field, California.

Quote :

"People have found a pimple here and they are trying to make it into a mountain," he told New Scientist. "A lot of the planets will show up regardless."

This is not the question. The question has to do with the detector sensitivity in auto mode and rocky planets due to the flickering.

Quote :

The team is developing software to automatically remove the noise from
data after it is sent down to Earth, but rigorous testing means it will
not be ready until 2011.

So the data will be fixed on Earth? The device is unfixable?

Quote :

The delays would only affect habitable planets around smaller, cooler
stars. The habitable zone for these stars is closer in, where planets
could complete the necessary three orbits in about one Earth year.

If anything these should be easier to detect. Or is he saying the new software can fix data being collected right now which does not have any flags in it.

TheoA, I have to say I think that you're overhyping this a little bit. Only 3 of the 84 amplifiers are degraded - they're still going to get lots of good science from the rest of the spacecraft's field of view. Also, remember that because of the quarterly roll the four sets of stars effected only have degraded photometry for 1/4 of the year.

Which, incidentally, is why Borucki said that this will preferentially effect Earth-anlogs around M-dwarfs. The short orbital period means that this 1/4 of a year degradation will hit their transits the hardest.

Interesting article

Quote :

Still, the first discovery of a habitable Earth-size exoplanet could be delayed if it happens to orbit a star monitored by one of the noisy detectors, Borucki admits.

Quote :

Without the glitch, this kind of planet could in principle be confirmed in 2010. But if observations of the planet were on a noisy CCD, that confirmation would be postponed until the noise-correcting software was ready in 2011.

I.e. data do not mix between different CCD?

Quote :

There is some good news. The stars Kepler is staring at are less intrinsically variable than had been predicted, which should make it easier to see transiting planets. "Most of these stars aren't as variable as we expected and that's going to help us," Borucki says.

The first good unplanned news after article in Nature.

It is very interesting that certain commentators in this thread have been taking the viewpoint that everything the Kepler team are saying is lies and spin, while trusting only in the media reports.

In particular, the Russian article is incorrect that the data is mixed between the CCDs on the satellite before the data is sent to the ground. The data processing on the satellite is extracting the data from the readout buffers and extracting pixels of interest, followed by data compression.

The data is thus processed on the ground, thus the ability to flag noisy data. It also means that the data can be re-reduced once the new algorithms are developed (a process that is expected to take about a month or so). Kepler is not limited to hot Jupiters, there is no sudden degradation in going to automatic mode. Sure, not having such great reliability of some of the detectors is going to reduce the sample of detected planets, but this is not a global problem that affects all of the data as has been claimed upthread.

It is nevertheless interesting that certain commentators in this thread have immediately leapt to the paranoid hypothesis that everything the Kepler team are saying is "spin". For example, it was remarked upthread that the "casual tone of the principal investigator" was cause for concern. Seems more likely that whatever "casual tone" was involved is because they are aware it is nowhere near being a showstopping problem. Do you expect them to run around claiming the sky is falling when in fact it is not, just to give media drama?

The paranoid attitude continues through the thread with every claim that the problems with Kepler were unrecoverable being treated as gospel and every claim made by the Kepler team to the contrary being considered "spin".

Frankly this is ridiculous. It is interesting to note that the general tone of discussion on Extrasolar Visions II has been generally hostile towards the experiments being done to detect extrasolar planets. I thought we'd seen the worst of it with the complaints about the long intervals between getting the data about the CoRoT planets, but this thread has really set a new low.

Maybe the "information instantly" expectations of the web don't mesh well with the messy realities of experimental science. I don't know.

Regrettably I won't be able to post any direct quotes. However I believe the summary I have given covers the points but bear in mind that I may have misinterpreted or introduced inaccuracies - these are my own fault and not due to the Kepler team.

Lazarus wrote:

In particular, the Russian article is incorrect that the data is mixed between the CCDs on the satellite before the data is sent to the ground. The data processing on the satellite is extracting the data from the readout buffers and extracting pixels of interest, followed by data compression.

The data is thus processed on the ground, thus the ability to flag noisy data. It also means that the data can be re-reduced once the new algorithms are developed (a process that is expected to take about a month or so). Kepler is not limited to hot Jupiters, there is no sudden degradation in going to automatic mode. Sure, not having such great reliability of some of the detectors is going to reduce the sample of detected planets, but this is not a global problem that affects all of the data as has been claimed upthread.

Thank Lazarus for information!

Can write the author of the article on www.Infox.ru disclaimer? Just take as its basis - the text of your message through a letter from the Keplet project participants?

In principle if people with project Keplera can show the offensive messages from this forum possible their delete? I sorry that believed the journalist.

Lazarus are you saying the common mode information is somehow kept track of and transmitted to Earth so the full signal can be recreated.

If so the errors could potentially be corrected on the ground.

Also the Kepler Website has omitted to mention this.

The way it reads right now it is easy to surmise that the common mode is permanently lost and only the differential light curve is transmitted to Earth.

It is good to hear there is no degradation in Auto mode but Borucki has not said this yet explicitly.

Excellent news if this not a global problem.

I apologize for the earlier comments if they offended anyone.

We wouldn't be here if we didn't care.

I'm just a little more skeptical of the news the team releases now.

TheoA wrote:

Lazarus are you saying the common mode information is somehow kept track of and transmitted to Earth so the full signal can be recreated.

If so the errors could potentially be corrected on the ground.

Apparently the information (the number of loaded photons for each pixel) before the passage of the amplifier is also recorded?

TheoA wrote:

Lazarus are you saying the common mode information is somehow kept track of and transmitted to Earth so the full signal can be recreated.

Not to preempt Lazarus, but there is no data reduction that occurs on the spacecraft. Kepler transmits little postage stamp images that are reduced by the science team on the ground. So all of the information is still there. Note that CoRoT does do onboard data-reduction, which is why they haven't been able to effectively deal with a lot of their systematics.

Also, the amplifiers govern the electrons -> digital number step of data collection, so it is not something that can be easily de-trended or otherwise corrected for on the ground.

Basically the questions I asked were about how much data processing was done on satellite vs on ground and whether re-reduction was possible. This was because these are the core points of the Russian article which was previously being taken as the most accurate (presumably because it was the most pessimistic, thus fed into the world-view of those who wanted to believe that Kepler was completely broken and the scientists were lying).

The information is that the data is processed on the ground, allowing for flagging and manual processing (contrary to what the pessimistic articles stated), and that re-reduction with improved algorithms is possible. Thus Kepler still has capability of detecting Earth-size transits even with the current version of the software. Of course, detecting them in the noisy channels is another matter.

As I suggested upthread, it is probably more worthwhile contacting the Kepler people yourself with questions than it is to go asking here.

SpaceCowboy wrote:

Also, the amplifiers govern the electrons -> digital number step of data collection, so it is not something that can be easily de-trended or otherwise corrected for on the ground.

Agreed.

The 15% area affected in turn will cause problems that may be uncorrectable.

Our fear was that this data was then used DURING the common mode rejection tainting the rest of the light curves.

Haven't asked about the affected area, but as far as I can see it, while 3 amplifiers failed leads to roughly 14.3% of the area being affected once rotation is accounted for, this area is not continuously affected, but within the affected area the noisy data would occur 25% of the time due to the quarter rolls.

This is not the same as a 14.3% loss of data. Unfortunate for planets with orbits that lead them to be synchronised with the noisy amplifier covering that region of the sky, but not a complete loss in the affected area.

Thank you Lazarus for taking the time and sorting this out. Kepler is a fantastic and difficult experiment. Your post made my day !