In order to better represent the C&R for the final catalog
, a prefilter was used to select
sources that fell within the catalog selection criteria. The prefilter was based on the
selection criteria used for the last catalog draw and performed on
each of the 11 scans covering tile 315825. Then the C&R analysis was executed.
Prefiltered C&R plot for tile 315825
Prefiltered J band C&R table for tile 315825
Prefiltered H band C&R table for tile 315825
Prefiltered K band C&R table for tile 315825
Next, the selection removed all single band sources prior to prefiltering. This
resulted in a C&R result for sources that had multiband detections only.
Multiband C&R plot for tile 315825
Multiband J band C&R table for tile 315825
Multiband H band C&R table for tile 315825
Multiband K band C&R table for tile 315825
Finally, only single band sources were used.
Singleband J band C&R table for tile 315825
Singleband H band C&R table for tile 315825
Singleband K band C&R table for tile 315825
There is no plot for the singleband sources since the C&R is zero for all
magnitude bins.
To summarize, I have plotted the completeness and reliability numbers for
the three populations of sources outlined above. There are two plots per band:
one showing the completeness numbers computed for each magnitude bin and the
other showing the reliability numbers computed for each magnitude bin. In the
plots each population of sources is represented with a different symbol. The black asterick shows the
numbers for the original, unfiltered C&R results; the red triangle
shows the numbers for the prefiltered sources; and the blue circle shows the
number for the multiband and prefiltered sources.
Completenes summary plot for J band
Reliability summary plot for J band
Completenes summary plot for H band
Reliability summary plot for H band
Completenes summary plot for K band
Reliability summary plot for K band
Date: October 19, 2001
******************************************
******************************************
C&R analysis was run after fixing some bugs in SMERGE that were introduced when SMERGE was modified to remove all apparitions of a source if one apparition had at least one indicator suggesting upl/confusion/artifact/persistence in any band.
Notice that the completeness in J band is now acceptable, as well as in H and k bands. The reliability drops slightly from the requirements due to extended sources in J band and satellite blips in H and K bands.
All sources within 10" of inner-most edge were also removed from the analysis.
Tiles 315824, 315825 , 315826 and 315822 were chosen due to repeated scan coverage, 12, 11, 8 and 8 respectively. All are southern hemisphere tiles.
The following analysis was done on tile 315825 only. Results show that the reliability is affected by extended sources and possible satellite detections in J and H respectively.
|
|
|
|
|
|
|
|---|---|---|---|---|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Explanation of categories:
Extended - visual check confirms source is extended
Satellite - visual check identifies a rectangular speck, clearly not a source.
This could be a satellite, or residual speck from a meteor streak. It appears mainly in H band.
Split - This occurs when more than one detection exists in a band for a single source OR
a single detection has associated magnitude different than 1 magnitude from the other
detections. This can cause a source to fall in the 1 times confirmed column as well as
the 10 times confirmed column (assuming 11 times confirmed is max).
Confused - This occurs when detection is on a defraction spike from a source that is
off the scan edge, or clearly within the confusion of another source. This seemed to
occur most in K band.
Odd - This specific source had two detections in a single scan but the source did not merge into
a single source. I put this into a category by itself until I understand why
bandmerge did not join these detections.
Comments and feedback: Sherry Wheelock
