Changes between Version 81 and Version 82 of MD10.refstack.20130715

Timestamp:

Sep 13, 2013, 2:18:16 PM (13 years ago)

Author:

Mark Huber

Comment:

--

MD10.refstack.20130715

@@ -401 +401 @@
 
 
+How many nights have been missed due to <4 exposures?
+{{{
+| dateobs             | comment                      | count(exp_name) |
++---------------------+------------------------------+-----------------+
+| 2008-09-18 11:47:48 | md10 test exposure           |               1 | 
+| 2008-10-09 10:14:56 | md10_2008-10-08 MD10 center  |               2 | 
+| 2009-12-27 05:35:05 | MD10 i MJD55082 MD10 center  |               1 | 
+| 2010-09-19 08:10:45 | MD10 g,r N5458 r MD10 center |               2 | 
+| 2011-08-28 14:12:18 | MD10 y N5800 MD10 center     |               3 | 
+| 2011-10-28 06:11:09 | MD10 y N5862 MD10 center     |               2 | 
+| 2011-11-20 06:17:02 | MD10 i N5885 MD10 center     |               1 | 
+| 2012-08-17 13:07:29 | MD10 y N6156 MD10 center     |               2 | 
+| 2012-11-09 05:13:26 | MD10 y N6240 MD10 center     |               3 | 
+| 2012-11-30 04:33:23 | MD10 y N6261 MD10 center     |               3 | 
+| 2012-12-11 05:48:44 | MD10 r,i N6272 i MD10 center |               2 | 
+--> so 11/444 nights, ~2.4% -- other fields similar ~1.4--3.6%
+}}}
+
+
+
 == Test 1D Night Stacks  ==
 
@@ -584 +604 @@
 [[Image(o5503g0240o.s045mk_cr_center.jpg,500px)]]
 
+----
 
 === PV2 Update Night Stacks ===
@@ -712 +733 @@
 Depth comparison -- 
 
-
+----
 
 === PV2 Update Night Stack SSdiff ===
@@ -795 +816 @@
 
 
-
+=== PV2 Night Stack Summary ===
+
+Was at least a good test to help flush out issues with the 1D convolution method. Basically want to decide if there is an improvement without a decrease in robustness, maybe also use for reprocessing only. 
+ * hard cut for FWHM ~12 to match the camera stage rejection does not quite work, extra allowance needs to be made for edges by ~2 pixels. The functional method used for making the LAP stacks may also be an alternative.
+ * 1D convolution appears equivalent in terms of the unconvolved stacks, and more robust at not 
+ * Nin minimum 2 -- useful in filling in regions or actually producing stacks in nights with fewer inputs
+ * SAFE false (single input pixels allowed) is somewhat uncertain -- more area on edges is definitely seen, how useful it is as a single warp is the question versus just using the warps for them.
+
+
+----
 
 = MD+3PI stack =
-Gene has suggested including overlapping 3PI exposures may improve the edges of the deeper stacks. A test sample run has been made with MD10 and 1D convolution for comparison to the pure MD10 refstack.
+Gene has suggested including overlapping 3PI exposures may improve the edges of the deeper stacks. A test sample run has been made with MD10 and 1D convolution for comparison to the pure MD10 refstack (would really only be used for the deep stacks, but refstacks are faster and easier to produce).
 
  * label, data_group -- MD10_3PI.teststack.20130728_1dg, MD10_3PI.teststack.20130728 (need to fix this)
@@ -804 +834 @@
 Like the refstack itself, 2009 data is not used and only the exposures from 5/2010--6/2013 were used -- this reduced the number of input warps from ~22-->12
  
- 
+
+Edge image samples --
+|| only MD || MD+3PI ||
+|| [[Image(.jpg,500px)]] || [[Image(.jpg,500px)]] ||
+
+
+Detection comparisons --
+
+Photometry comparisons -- 
+
+
+
+== MD+3PI Stack Summary ==
+
+
+