Multiframe comparisons between v4 (with old dynacal) and v6 (new dynacal) _________________________________________________________________________ Notes: * all plots below pertain to multiframe extractions from tile 3306m561_wppm41 * fops = processed with v4 with frames calibrated using dynacal version that introduced a "false mode" in pixel distributions. * As a reminder, this "false mode" is one of the causes for the "big" photometric bias seen in the All-Sky Source Catalog. * pops = processed with v6 where frames were reprocessed using new dynacal software that corrected for the "false mode", and cleaned up some other bugs, e.g., over-masking of bad (transient pixels). * filtering performed in fops and pops for band of interest: 0.5<w?rchi2<2.3; no upper limits; na=0; nb=1. * goal: are there significant differences between source metrics (e.g., photometric uncertainties) between these versions to warrant recreating frame masks and uncertainty products when reprocessing cryo data with the fixed (v6) dynacal? * conclusion (answer to previous question): there does not seem to be a significant difference in source metrics. Plots: In the following, note the flux-underestimation bias in fops (v4) relative to pops (v6). Vertical lines are s/n=5 (multiframe) limits for v6 for this tile, deduced from the w?sigmpro vs mag plots below. W1mpro_fops_vs_pops W2mpro_fops_vs_pops W3mpro_fops_vs_pops W4mpro_fops_vs_pops The following show differences in the single-exposure RMS repeatability. The rational here is that if more pixels were masked in fops (v4), then fewer pixels were being used in profile-fit photometry in general. This should manifest itself as an inflation in the w?sigP1 values - i.e., more uncertain single-exposure photometry in fops. This does not appear to be the case. Vertical lines are s/n=5 (single-exposure) limits for v6 for this tile, deduced from the w?sigP1 vs mag plots below. W1sigP1_fops_vs_pops W2sigP1_fops_vs_pops W3sigP1_fops_vs_pops W4sigP1_fops_vs_pops Same as previous set, but as scatter plots: W1sigP1_fops_vs_pops W2sigP1_fops_vs_pops W3sigP1_fops_vs_pops W4sigP1_fops_vs_pops Vertical lines in following plots are s/n=5 (multiframe) limits for v6, for this tile, deduced from the w?sigmpro vs mag plots below. Note the upturn towards faint magnitudes. This is a consequence of the flux-underestimation bias in fops (v4) since w?sigmpro ~ sigma(flux)/flux and the action here is all in the flux, not sigma(flux). Hence for a given (faint) magnitude, the reported photometric (mag) uncertainties are overestimated in the All-Sky release. W1sigmpro_fops_vs_pops W2sigmpro_fops_vs_pops W3sigmpro_fops_vs_pops W4sigmpro_fops_vs_pops Same as previous set, but as scatter plots: W1sigmpro_fops_vs_pops W2sigmpro_fops_vs_pops W3sigmpro_fops_vs_pops W4sigmpro_fops_vs_pops Following were used to estimate approximate s/n=5 (multiframe) limits for v6 for this tile (vertical lines). Assumption here is that the "model-based" w?sigmpro values reflect the actual photometric uncertainty. W1sigmpro_pops_vs_mag W2sigmpro_pops_vs_mag W3sigmpro_pops_vs_mag W4sigmpro_pops_vs_mag Following were used to estimate approximate s/n=5 (single-exposure) limits for v6 for this tile (vertical lines). These are based on the single-exposure RMS repeatability and are not sensitive to confusion noise like the w?sigmpro values. W1sigP1_pops_vs_mag W2sigP1_pops_vs_mag W3sigP1_pops_vs_mag W4sigP1_pops_vs_mag -- F. Masci, 12/13/2012