search

davidwhogg / SloanAtlas

code to assemble the plates of the Sloan Atlas of Galaxies
4 stars 0 forks source link

inspecting geo mean outliers #77

Open ekta1224 opened 10 years ago

ekta1224 commented 10 years ago

here are the old(black) versus new(magenta) h50 and h90 values: http://broiler.astrometry.net/~ep1091/check_geomean.pdf

some flipbooks for the outliers: http://broiler.astrometry.net/~ep1091/all_flipbooks/flip-NGC_7678.pdf http://broiler.astrometry.net/~ep1091/all_flipbooks/flip-UGC_10245.pdf http://broiler.astrometry.net/~ep1091/all_flipbooks/flip-NGC_7556.pdf http://broiler.astrometry.net/~ep1091/all_flipbooks/flip-NGC_4567.pdf http://broiler.astrometry.net/~ep1091/all_flipbooks/flip-UGC_2728.pdf

i don't really see a relation between these. any ideas on what might be throwing them off? my guess is that their jpegs could be crappy

ekta1224 commented 10 years ago

i've also looked at objects with the greatest np.abs(delta(h90/h50)). the top five are:

NGC 4222, 5.281 (http://broiler.astrometry.net/~ep1091/all_flipbooks/flip-NGC_4222.pdf) IC 1515, 5.776 (http://broiler.astrometry.net/~ep1091/all_flipbooks/flip-IC_1515.pdf) A0200+21, 5.7836 (http://broiler.astrometry.net/~ep1091/all_flipbooks/flip-A_0200+21.pdf) UGC 6594, 15.446 (http://broiler.astrometry.net/~ep1091/all_flipbooks/flip-UGC_6594.pdf) UGC 4148, 18.312 (http://broiler.astrometry.net/~ep1091/all_flipbooks/flip-UGC_4148.pdf)

davidwhogg commented 10 years ago

uh oh 2728 is totally wrong! What should we do about that one? It needs the star masked (it hasn't been).

ekta1224 commented 10 years ago

ok, i can re-initialize that one alone.

ekta1224 commented 10 years ago

any ideas on the others?

davidwhogg commented 10 years ago

not yet; keep pestering me. Do you have any ideas on them?

ekta1224 commented 10 years ago

i pulled out the gri images for the outliers besides UGC 2728: http://broiler.astrometry.net/~ep1091/quantiles/quantile06/NGC_7678_SA_J232827.91+222516.2_irg.jpg http://broiler.astrometry.net/~ep1091/quantiles/quantile14/UGC_10245_SA_J160948.05+490859.7_irg.jpg http://broiler.astrometry.net/~ep1091/quantiles/quantile14/NGC_7556_SA_J231544.46-022253.4_irg.jpg http://broiler.astrometry.net/~ep1091/quantiles/quantile12/NGC_3640_SA_J112106.84+031405.8_irg.jpg http://broiler.astrometry.net/~ep1091/quantiles/quantile10/NGC_4567_SA_J123633.18+111523.3_irg.jpg

could the bad images for the first, third, and fourth above have something to do with it? i'm not sure what the direct correlation is between the i-band fits files used in the halflight measurement and the gri images shown here but that's the only guess i have right now.

ekta1224 commented 10 years ago

p.s. those 3 that i pointed out with bad images are the farthest from the line on the outliers plot too

davidwhogg commented 10 years ago

Yes the bad images would do it!

ekta1224 commented 10 years ago

Sweet! Is there any way to fix those? Rather, do we want to worry about fixing those?

On May 7, 2014, at 6:47 PM, "David W. Hogg" notifications@github.com wrote:

Yes the bad images would do it!

— Reply to this email directly or view it on GitHubhttps://github.com/davidwhogg/SloanAtlas/issues/77#issuecomment-42493061 .

davidwhogg commented 10 years ago

well, we can try re-building. I am not sure why the tractor gets more images than the image-building code gets.

ekta1224 commented 10 years ago

okay, i will redo a few of them and see if we get anything better. seems odd to me too.

On Thu, May 8, 2014 at 10:04 PM, David W. Hogg notifications@github.comwrote:

well, we can try re-building. I am not sure why the tractor gets more images than the image-building code gets.

— Reply to this email directly or view it on GitHubhttps://github.com/davidwhogg/SloanAtlas/issues/77#issuecomment-42627038 .

Ekta Patel