Ask an expert

Existing data

• In catdir.20090327, i seems to be nan throughout, can that be right? or did I look in a region that has no i-band imaging? How would I tell? (also see questions on dvo catalogs below)

• Similarly, in catdir.ThreePi.Run2.v0, the flag column seems to be all 0? (Also see more flag-related questions below)

• Eric Bell suggested: for synth.grizy calibration, compute only one ZP per FPA not per chip - too few blue stars to caibrate on per chip. Calibrate chip-to-chip variations via flatfield only. (For uebercal with relphot, it's a different story!)

dvo (inc. addstar, relphot, uniphot, relastro, dvo shell)

dvo catalogs & their generation, processing/stacking strategies etc.

• Are the errors returned by avextract r:err etc. the RMS of the measurement, or the suitably averaged input errors, or ...? How are the averages computed - any outlier clipping applied? weighted mean? ... ?

Answer from Mark Pitts: To the best of my knowledge, the suffix "err" when used in avextract refers to the error of the mean (i.e., the error on r:ave). I'm afraid I don't know the details of how the average is computed.

• How do I make a lightcurve for a given object?

Answer from Mark Pitts:

1) You can use "mextract" to pull individual measurements, including magnitude (using the proper photcode) and time of the measurment (using the "time" keyword). Eventually, you will be able to use the "OBJID" keyword to pull all the measurements associated with a given object, but for now you can use the RA and DEC keywords to isolate the object (i.e., use boolean logic to pull out all measurments within about 1 arcsec of the object):

-> catdir (data catalog)

-> region (ra) (dec) (radius)

-> define values for object's position obj_ra and obj_dec

-> mextract -region (photcode),(photcode):err,time where (sqrt(((RA-obj_ra)*cos(DEC))^{2 + (DEC-obj_dec)}2) <= 0.0003)

2) Now you can plot up the data:

-> set xvec = time; set yvec = (photcode); set zvec = (photcode):err

-> device -n lightcurve

-> labels -fn times 14 -x "time (sec)" -y "mag" +x "Lightcurve"

-> limits xvec yvec; box -lw 2

To plot points:

-> plot -x 2 -pt 1 -sz 1.5 -c black xvec yvec -dy zvec

To connect the points:

-> plot -x 0 -lt 0 -lw 1 -c green xvec yvec

• We want to find dropout objects, which may involve identifying objects in one filter (y, say) and then demanding non-detection in another one or more (iz, say). How would I run a dvo query to find all y detections without iz detections?

• The observing plan includes 2 images per field per night. Are these stacked immediately, and objects detected on the stack? Or only on individual images?

Answer (Gene): The current strategy doesn't have a stack of the 2-way images, just chip cam fake warp, then two pair-wise diffs (both "forwards" and "backwards") to allow transient detections and magic to happen. Doing a per-night stack for all filters is probably too time-consuming, but might be worth considering for y given that its throughput is the lowest and it's observed closest to the sun, and with the largest separation in time between every year's 2 visits to the same part of the sky.

• Would it be feasible (space- and processing time-wise) to drop the default 3pi detection significance threshold to 3-sigma, say, to allow automatic catalog-level stacks (yielding ~5-6-sigma catalog detections from a year's four individual visits)?

• In general, what is the plan for making stacks - at fixed calendar dates, whenever a given region of sky has become unobservable for the year, ...?

• At which processing stage, and on which timescales, will forced photometry be run in all "other" bands based on detections in one particular band? Do we have to do forced photometry ourselves to get griz fluxes for y-only detections within a few days or weeks?

• What happens to negative fluxes (which will appear in forced photometry) when they get converted to magnitudes? Luptitudes have the correct behaviour for colours, but we have already decided that they are not practical in a collection of surveys with different and evolving flux limits, such as PS1SC 3pi + MDS + Andromeda. How do I do a catalog search for flux ratios exceeding some threshold (i.e. a colour cut) if I don't have luptitudes?

• What is the default stack optimized for? Is there a preconfigured way (assuming the need) to do stacks optimized for other things, e.g. for faint point-source photometry, for good seing only, ...?
  • Answer from Gene: The default stack convolves all the input images to something slightly worse than the worst seeing.

dvo shell

• How do I find out which filters/instrumental photcodes are present, and to which photcodes they can be translated, in a given catdir?

Answer from Mark Pitts: The photcodes can be examined by exporting the photcode table:

1) Export photcode table from catdir to a file:

photcode-table -D CATDIR (catalog) -export (filename)

2) Examine and/or edit the file as needed

3) Import the edited file back into the catdir:

photcode-table -D CATDIR (catalog) -import (filename)

Note: the values in the catdir are not automatically updated if the photcode table is changed. You may have to rerun photometry commands (such as relphot)

• How do I find out the exposure time for a given object, and/or the "effective" exposure time of a given average magnitude?

Answer from Mark Pitts: The keyword "exptime" can be used in mextract to pull out the exposure times for individual measurments. I'm not certain if there is an easy way to pull the "effective: exposure time of an average magnitude.

• When I plot image outlines with 'images' and/or use gimages to find images overlapping a given coordinate, again how do I tell the filter?

• How do I extract only objects from a given exposure or chip within an exposure?

• What is supposed to happen when doing multiple avextract commands with different regions? I'm asking because we tried many different avextract commands with different regions, and somehow ended up with (allegedly) 30,000 objects in (0.001 degree)² around (14.84438,0.05039). But maybe it was just because...

• ... the difference between region and skyregion is a bit subtle, and it's not obvious which out of avextract -region, avextract -skyregion ra1 ra2 dec1 dec2 and "naked" avextract is what one wants. I think it's because the names don't make clear that one refers to the plot window and the other one sets the region of interest for avextract etc. How about these suggestions:
  • Rename region to plotregion, and instead of `avextract -region, use avextract -useplotregion` or similar
  • Rename skyregion to extr_region (if one does that to use with later avextract commands, one won't have to type it very often per dvo session, so a long name is ok).
  • Analogously, use avextract -extr_region ra1 ra2 dec1 dec2
  • In any case, have avextract print out which region it used, to give a visual check for the dvo user that s/he's doing what was intended.

• When using avextract ... flag, for which band do I get the "flag" data? If the flag is not just for a single measurement , how is the "object" flag computed from the "measurement" flag?
  • It would be useful to have some sort of help text for the flags from within dvo, and in the long run also to build in a simple way to apply some set of flag cuts that are useful to restrict oneself to reliable point sources, reliable galaxy magnitudes, etc.

• How do I open and address additional plot windows, e.g. to have one for an image, one for CMDs, and one for plotting objects in sky coordinates?

Answer from Mark Pitts: To open new graphics windows:

device -n (window name)

Without specifying a name, DVO with assign them numbers starting with zero. You also use this command to "activate" the window to plot or display stuff.

• Assuming a local image mirror, we would like to do things like:

• Plot a CMD, click on an object in the CMD and have its extracted averages and/or measurements highlighted, or a light curve generated, or the postage stamp pop up, the entire image loaded in a tv display with all objects overlaid. In general, have some action taken when clicking on a plot window or tv display window that has objects near the cursor position.

• region -30 20 90 AIT leads to segfault? In any case, AIT, centered on not 0,0 gives weird plot results

Answer from Mark Pitts: The region command accepts the position options as:

region (central RA) (central DEC) (radius); all units are in degrees

It may be that using a negative RA values caused the problem.

• How do I make axis labels?

Answer from Mark Pitts: Using the labels command:

labels -x "x axis" -y "y axis"

This command can also add a title and place strings in the corners of the plot area.

ppImage

• How do I run ppImage from the command line with the PPIMAGE_OA_SOFT recipe from ippconfig/sdssmosaic/ppImage.config (which turns all steps off except subtracting the constant software bias of 1000 counts via OVERSCAN)? We tried

   ppImage ... -recipe ...
   # and
   ppImage ... -F SOMETHING SOME_PPIMAGE_RECIPE
  

but it complains about not having been given a bias frame, even though BIAS = FALSE in the recipe.

psphot

• How is deblending of overlapping objects done?

• How is star/galaxy separation done?

• How do I get aperture mags, fluxes instead of magnitudes, etc. in the output .cmf table - i.e. more or different outputs?

• How do I feed an external catalog to psphot for forced photometry?

• what's PEAK_FLUX_AS_MAG in the .cmf file?

Answer from Michael Wood-Vasey: -2.5*log10(counts in peak pixel). The counts in the peak pixel expressed in instrumental magnitude.

• Is there a tool for doing quick overlays of the .cmf table on the input image?

pswarp

ppStack

• Is there any rejection of cosmic rays based on outlier rejection in stacking? Or does that mess up moving & variable object photometry?

ppSub

panTasks

• Where are the command lines stored for the individual analysis commands that pantasks launches (e.g. the ppImage call for some image I read about in ippMonitor)?

ippMonitor