The KIDS survey

Intro: what is KIDS

• KIDS, the KIlo-Degree Survey, is a 1500 square degree public imaging survey in the Sloan colors (u',g',r',i',z') with patches in both the Northern and Southern skies. The survey will use the OmegaCAM instrument mounted on the VST (VLT Survey Telescope). It will start summer 2011.
• A parallel survey called VIKING (VISTA Kilo-degree INfrared Galaxy survey) is taking place on the VISTA telescope on Cerro Paranal since February 2010. It covers the 1500 square degree of the VST-KIDS survey in the 5 broadband filters Z,Y,J,H,Ks. Combined with KIDS, this will yield a unique 9-band optical-IR survey with depth approximately 2 mag deeper than Sloan and 1.4 mag deeper than UKIDSS-LAS.

The proposal to ESO

Complete proposal text	(PDF)
ESOFORM template pt. 1	(PDF)
ESOFORM template pt. 2	(PDF)
ESOFORM template pt. 3	(PDF)
Management plan v.2006/4/15	(PDF)

The time allocation

Here's what has been approved (in hours):

Period	u,d,0.85-1.1	g,d,0.7-0.85	r,d,< 0.7	i,b,< 1.1	i-wide,b,< 1.1	calib,b,any
P77	60	60	119	143	51	43
P78	60	60	119	143	51	43
P79	60	60	119	143	51	21
P80	60	60	119	143	51	21
P81	60	60	119	143	51	21
P82	60	60	119	143	51	21
P83	60	60	119	143	51
P84	60	60	119	143	51
TOT	480	480	952	1144	408	170
GRAND TOT	3634
P85		120
P86		120
P87		120
P88		120
TOT REPEAT		480
GRAND TOT INC REPEAT	4114

The fields

KIDS will target two patches, KIDS-N and KIDS-S. The total area is 1500 sq.deg.

KIDS-N runs along the equator in the NGC, from RA 10:00 to 15:52 (150 to 238 deg), and between declination -5 and +3 (with the exception of RA> 15, dec< -3). In addition it includes the area RA 8:40 to 9:08, dec -3 to -1.

KIDS-S is a single rectangular patch around the SGP, between RA=22:00 and 3:30, and dec -35 to -25.

The location of the fields, plotted on top of the Schlegel, Finkbeiner, & Davis 1998, ApJ, 500, 525 reddening maps, is shown below. The areas are outlined in green; B-V extinction is contoured at intervals of 0.05.
Click on the images for a larger version.

KIDS-N:

KIDS-S:

Order in which fields are to be observed

To increase the science return from the survey before it is completed, we have considered the best order in which to build up the survey. Relevant considerations are

1. a desire to get a useful angular power spectrum on large scales
2. a wish to maximize coverage of the GAMA AAO spectroscopic survey patch contained in KIDS-N
3. not compromise schedulability
4. build in some overlaps of adjacent fields to allow calibration to be checked and improved

The following proposal emerged from this (discussions KK, Wilman, Phleps, Schuecker). It is illustrated here for KIDS-N. (Click for a larger PS version).

The KIDS-N patch is divided into square blocks, 9x9 degrees. A half-hour integration, plus 25% overheads, takes about 38 minutes, and in this time the sky rotates 9 degrees on the equator. So on a good night one field can be observed from each block in r, or two fields in g (or u), etc.
The blocks come in two varieties, 'inc' and 'dec' (increasing and decreasing RA coverage). They alternate along the KIDS patch.
Within a block the sequence of observation is as shown in the lower panels of the figures. The first two declination strips are observed in opposite directions in the two blocks, so that where the blocks meet contiguous areas build up rather quickly. After about 1/4 of the patch is observed we will have a complete 2 degree-wide strip. From this point on, the remaining strips can be built up in order of increasing RA, and the distinction between Inc and Dec is no longer important.
The first fours strips to be completed are those between declinations -2 and +2, where the spectroscopic AAO survey will take place.

In the KIDS-S patch (declination -30) the sky rotates a factor 0.87 more slowly, so here the natural width of a block in RA is 8, not 9 degrees.

Tiling strategy

To avoid holes in the KIDS images, observations will be dithered with offsets sufficiently large to bridge the inter-CCD gaps of OmegaCAM. We will use 5 dithered exposures per field, which results in an exposure map that looks like

where white indicates the area covered in all 5 exposures, and progressively darker shades indicate lower and lower total exposure time.

At the edges of the fields, the exposure level decreases from 5 to 0, in a staircase pattern with step width equal to the dither steps (25 arcsec in X, 85 arcsec in Y).

The adjacent fields should mesh with each other in a nice way. Two possibilities are illustrated below (click on figure for postscript version):

These panels show the exposure levels at the edges of two adjacent dithered images. The top panels show the exposure versus X, the lower panels versus Y (which has bigger steps). The horizontal axes are labelled in pixels (below) and arcminutes (above), and the center of the field is at (0,0).

Option A is illustrated on the left. The field centers are separated by the full extent of the mosaic, so that the combined exposure level in the overlap area, obtained by summing the exposures from both dither sets, is everywhere nicely equal to 5. This minimizes the number of exposures but leaves no redundancy. It also means that in the data analysis results from different OBs need to be processed together, otherwise every pointing is surrounded by a rim where the total exposure is only 60% of that elsewhere. The field centers are separated by 3673arcsec = 61.2arcmin in X, and 3700arcsec=61.7arcmin in Y. Option B is shown on the right. The field centers are closer together than in option A (2 offset steps=50arcsec in X, 170 arcsec in Y closer). 80% exposure can now be achieved without having to add up observations from different fields, which simplifies the analysis. Moreover, there is now some redundancy: around each OB we have a strip that is 2 steps (50-170arcsec) wide, one degree long, that has received 4 exposures in one OB and 3 in another. The price for the overlap is (obviously) an increased number of field centers. The field centers are separated by 3623arcsec = 60.4arcmin in X, and 3530arcsec=58.8arcmin in Y.

Because of the redundancy and operational simplicity, option B is preferred. The resulting exposure pattern from a 2x2 set of fields is then as illustrated below. The darkest shade of grey represents 3/5 of the total exposure time. The grey horizontal and vertical strips marked with a + are the overlap regions where 4 exposures from one field overlap with three exposures from the adjacent one.

+ +

Further notes on KIDS Tiling and PlateSystem (Edwin Valentijn 8 June 2006)

Case B has a minimum overlap between adjacent fields. In itself this is not a problem, as the OmegaCAM calibration plan forsees for each field an  independent photometric  and astrometric calibration.  However, some overlap is desired as an extra cross check on calibrations. Case B provides a few 100 pixels overlap with 1-3 dithers; and that is tight but just enough.

Here, we explore the possibility to adopt this scenario for a whole sky PlateSystem. ( see Tiling the Sky for OmegaCAM)


defs:
A  field of fieldsize_ra* fieldsize_dec has a fieldcenter.
A 5 point dither  points  with respect to this fieldcenter  has  2 times ++ and 2 times -- in steps ditherstep_dec and ditherstep_ra.

In fact for a 5 point dither the transitions at the edges of the fields will have number of frames
   ...... 5 5 6 7 7 7 7 6 5 5 ......... per ditherstep.
The central two 7 dither strips are composed of 3+ 4 and 4 +3  dithers of the two fieldcenters.
Thus these two   strips  have a minimum of 3 dithers per fieldcenter, which is what we define the overlap.


When designing  a grid for a whole hemisphere this condition is kept as a strict  minimum.
In practice, several effects cause a larger overlap than this minimum:

• in RA and DEC: putting an integer number of fields in a RA strip, and on the meridian
  
• in RA: the cos (delta_dec) term over the 1 degree fieldsize_dec, which is larger than 200 px at dec <-45 at the top of an image compared to the bottom..
• in RA and DEC:  when ADC is in, the scale at the focal plane changes from 0.2134 to 0.2149 arcsec/pixel, enlarging the  fieldsize_ra from 1.01270 to 1.01982 degrees, and the fieldsize_dec from 1.02451  to 1.02738  degrees, which implies  an extra overlap of 112 px  in both directions.
  
• In Dec: differential atmospheric refraction enlarges fieldsize_dec
  

Case B also has the property that the average separation between fields in degrees is  larger in Dec than in RA direction because of the large horizontal gaps between the CCDs ( 100 px in Ra= ditherstep_RA,  374 px in Dec= ditherstep_Dec).

For the two-lens corrector:
fieldsize_ra =  1.01270 degree
fieldsize_dec= 1.02451 degree

With these setttings PlateSystem instantiates a Grid::


With this grid, PlateSystem  produces the following plots of the fields  for some Survey areas
(rectangles depict the actual field size of  OmegaCAM  and overlap corresponds to at least two dithersteps)

I have included the anticipated survey areas for other surveys, like VESUVIO-Horlogium and VESUVIO-Hercules.
PlateSystem will really pay off when we can convince the various surveys  to adopt  the same tiling (common archive, data reductions-overlap handling!). Note that  2 hours/field  propagates to  5*300 = 1500 field/year  or 15000 fields/10 year.
So when only half of the anticipated lifetime of OmegaCAM  will be devoetd to this kind of survey work 7500 fields can end  up into the archive, which is about  36% of  the southern Sky!  So PlateSystem could become a power tool for our quest for larger area.




dd


dd


dd

dd

dd


ddd


ddd

dd