MEMO TO SPITZER OBSERVERS WITH EARLY RELEASE WARM IRAC DATA In this memo we discuss the most salient issues with data from early warm IRAC observations, and compare them to cryogenic IRAC data, to let observers know what the main changes are. Overall Processing of Data Warm IRAC data are processed using the same methodology as cold IRAC data. The main differences in processing are that there are no laboratory bias measurements to use and the muxbleed correction is not applied. For the current processing of IRAC campaigns PC-1 (28 Jul - 12 Aug) and PC-2 (13 Aug - 26 Aug), a first-frame calibration is not applied and the cryogenic linearity solution is applied. The skydark subtraction, flat-fielding and flux scaling all use calibration data derived from observations at the same operating setpoints as the data in PC-1 and PC-2. The calibration data are either from these campaigns or from the IRAC warm instrument characterization (IWIC) period which preceeded PC-1. Subsequent campaigns will have different calibrations and data properties, depending on the operating setpoints at the time of the campaign. 1. Absolute Calibration and Uncertainty The flux conversion values for channels 1 and 2 (3.6 and 4.5 microns) in PC-1 campaign are 0.1245 +/- 0.0042 and 0.1490 +/- 0.0059 MJy/sr per DN/s, respectively. These values indicate a througput decrease of ~14% and ~7% for FPAs 1 and 2 compared to cryogenic operations. The uncertainty in the measured flux of any source is a combination of photon noise of the source and background, readnoise of the observation, the propagated error of the data calibration steps including dark subtraction, linearity correction and flat-fielding, uncertainty in the pixel phase and location dependent photometric corrections as well as the uncertainty in the transfer of photometric truth through the flux conversion value. For high signal to noise, well dithered observations, the photon noise, readnoise, pixel phase and location dependent correction uncertainties are small compared to the uncertainty in the absolute calibration (3%) and pipeline processing. The uncertainty in the pipeline processing for bright sources is currently dominated by the linearity correction. We estimate the total uncertainty for bright sources to be 5%-7% in channel 1 and ~4% in channel 2. 2. Photometric Stability The photometric stability of IRAC has been measured using both a 36-hour sequence of standard skydark and calibration star observations and a set of high precison photometric monitoring observations. An initial analysis shows that the warm IRAC exhibits the same photometric stability as in the cryogenic mission. The standard calibration star measurements do not vary in time at a level of 2% or more. The high precision photometry observations were able to recover about 60% of the cryogenic Poisson limit which compares quite favorably with cryogenic observations which reached 70% of the Poisson limit. High precision photometric observations are limited in precision by the ability to remove the pixel-phase effect and the intrinsic Fowler sampling of the observations. There are no indications that residual images cause a time variability in the observed light curves. 3. Optical Properties of the Arrays The optical properties of the telescope and instrument have not changed measurably since the cryogenic mission. The array locations and orientations are the same to within 0.1" in position and 0.1 degrees in rotation. Measurements of the array distortion also indicate no significant changes from the cryogenic mission to within measurement uncertainties ranging from 0.1 pixels in the array centers up to ~ 0.5 pixels in the array corners. Focus measurements also indicate that the focus has remained close to the cryogenic mission value. We expect to be able to quantify this better once the linearity correction has been improved. 4. IRAC Dark Frames Since IRAC does not use a photon shutter for a dark measure, a pre-selected region of low zodiacal background in the north ecliptic cap is observed to create a "skydark." It is a measure of the bias unaccounted for in the lab measurements. During each campaign, a library of skydarks of all frametimes is observed, reduced, and turned into skyframes with the pipeline. For the warm mission, skydark suites are taken, reduced, and used in the same manner as in the cryogenic mission. Overall, a higher bias level is seen in the skydarks relative to cryo. In addition, clusters of bad pixels at 4.5 microns result in a series of artifacts (notably column pulldown) being present in every frame. These are in every frame taken, therefore they subtract out in BCDs. Median counts of the bias frames have increased significantly from ~ 30 DN/s in cryogenic mission to ~ 100 DN/s in warm mission in channel 1, and from ~ 3 DN/s in cryogenic mission to ~ 12 DN/s in warm mission in channel 2. URL: Link to skydark images during cryogenic and warm missions http://ssc.spitzer.caltech.edu/irac/img/darks.jpg The figure linked above shows the IRAC 12 second skydark ensembles created in cryogenic mission (left) and during the warm mission characterization period (right). The column pulldown artifact is apparent in five columns of the 4.5 micron channel in the warm mission darks. Its strength is the same percentage of the background for all the frame times. 5. IRAC Flat Fields The pixel-to-pixel gain variations (commonly called the flatfield) were remeasured during IWIC. These measurements were made as during the cryogenic mission, using highly dithered imaging of the brightest parts of the zodiacal cloud. Overall, the flatfield has remained substantially unchanged relative to the cryogenic mission. There are, however, small gradients and other changes at the 1% - 3% level, so the two are not interchangeable. Similarly, the normalization of the flats has changed slightly, so for self-consistency the current flux calibration factors only apply to data processed with the new flats. All currently released data are flattened using the IWIC flat. The accuracy of the flat is 0.7% (1 sigma) at 3.6 microns and 0.3% at 4.5 microns. Data taken with the temperature floating (after August 12th) may have slightly degraded performance. However, given the very small differences in the flatfield between 15 K and 30 K, it is likely that the flatfield is still accurate to better than 1%. URL: Link to channel 1 flat field images, showing a comparison between Warm and Cryo flat fields: http://ssc.spitzer.caltech.edu/irac/img/4419-ch1-embed.jpg URL: Link to channel 2 flat field images, showing a comparison between Warm and Cryo flat fields: http://ssc.spitzer.caltech.edu/irac/img/4420-ch2-embed.jpg 6. IRAC Linearity The IRAC detectors have a non-linear response - the conversion from detected flux to data numbers is not a simple constant. In the cryogenic mission, this non-linearity was corrected based on ground calibration of the instrument using special test equipment, and was confirmed in-flight through numerous observations. During IWIC, it was found that the non-linearity and well-depth varied as both a function of applied voltage on the arrays and the array temperature. Both of these have been changed as a result of warm operations. Unfortunately, it is extraordinarily difficult to recalibrate the linearization in flight with the accuracy of the ground-based cryogenic linearization. At the current time, all warm data are processed with the cryogenic calibration. IWIC results indicate that the non-linearity at warm temperatures is more severe than at 15 K. Using the cryogenic calibration at least provides a partial correction until such time as the warm calibration is completed and verified for use. Warm data linearized with the cryogenic calibration in channel 1 will have a systematic error of roughly 5%, such that brighter objects (those approaching the saturation limit near 30,000 DN) are systematically too faint. The effect is much smaller in channel 2, and is believed to be approximately 1%. Note that the flux calibration is based on standard stars which are usually near 1/3-1/2 full well. In addition to bright objects being too faint, faint objects may appear slightly brighter than they actually are, although this effect will be at the 1% level or less. The other change is that the well-depth which is defined by the point at which the pixel DN peaks is now roughly 30,000 DN in both channels, whereas it was closer to 44,000 DN in the cryogenic mission. Since the data are processed currently with the cryogenic calibration, the saturation flag should be considered unreliable. Users should examine the raw data numbers for sources that are suspected of being near saturation. 7. Persistent Images Similar to the cryo mission there are both short-term and longer term residual, or persistent images, in both channel 1 and channel 2, caused by exposure to a bright source in a previous image (See Section 4.3.8 in the IRAC Data Handbook). All observed residuals drop by a factor of 500-1000 in flux in the image directly after the offending bright source image. Short-term residuals last for a few minutes. Their decay time is related to the brightness of the source such that residuals from brighter sources will decay away faster than those from fainter sources. Residuals in channel 2 turn from positive flux to negative flux for exposure times less than 100 seconds. The longer term residual images are very different in behavior than residuals seen in the cryogenic mission. Instead of residual images which last for days or weeks unless annealed (the long-term residuals in channel 4 in the cryo mission), there are intermediate-term residual images which can last for several hours. Intermediate-term residual images seem to be flux-dependent such that intermediate brightness sources (~ 2nd -5th mag) leave the longest lasting residuals. These persistent images are different from those seen in the cryo mission in that they have flat flux profiles instead of the normal peaky profile. We do see "slew residuals" which appear as lines of residual flux after the telescope moves across bright sources. We have so far observed no week-long persistent images, similar to those that were seen in channel 4 during the cryo mission. We are not annealing the arrays as there is no evidence that annealing removes residual images, and all residual images decay in a reasonably short time scale (no more than a few hours). Our recommendation is to be aware of any bright sources in your data for which a factor of 500 drop in flux would still leave an image above the background level. A dithering pattern is very successful in removing contributions to a mosaic from residual images. A consequence of the intermediate-term residual images is that it is possible for previous observations to produce residual images in your data. Examining the median stack image is useful to help identify pixels affected by intermediate-term residual images. In principle, these residuals could be mitigated by subtracting the normalized median stack as a template for the residual image. 8. Other IRAC Artifacts In the cryogenic mission, two image artifacts were identified in the BCDs and corrected in the CBCDs, muxbleed and column pulldown. For a description of these effects, please see http://ssc.spitzer.caltech.edu/irac/features.html#2B and http://ssc.spitzer.caltech.edu/irac/features.html#2D or read the corresponding sections in the IRAC Data Handbook, 4.3.2 and 4.3.4. In the warm data, it appears that there is no muxbleed from sources, so it is no longer a problem. The column pulldown has been identified in the warm data, but the functional form of the artifact is different. The level of the pulldown above and below the triggering source is much different and the pulldown decreases with distance from the trigger along the column. A more complete characterization must be done before a full correction can be applied within the pipeline. 9. Array Location Dependent Photometric Correction Based on our analysis of observations of IRAC calibration stars, the location dependent photometric correction has changed from the cold IRAC correction. We are currently in the process of determining a new correction. For well-dithered data, the photometric variation should average out. Observations of sources towards the center of the array also have minimal corrections. 10. Pixel Phase Correction As in the cryogenic mission, there is a slight variation in flux depending on where a point source falls with respect to the center of a pixel ("pixel phase"). Generally, the photometric response is higher near the center of a pixel and lower near the edge, although the behavior is not perfectly radially symmetric, and not centered exactly on the middle of a pixel. In addition, the shape of the correction varies somewhat from pixel to pixel across a given array. We have analyzed data for the same standard star placed on or near the nominal pointing center in both the full- and sub-array fields of view (FOVs). In channel 1, the response varies by 7.4% across a pixel near the full-array FOV center and 8.1% near the sub-array FOV center. In channel 2, the response varies by 3.4% near the full-array FOV center and 2.1% near the sub-array FOV center. This contrasts with a 4% variation in channel 1 and a 1% variation in channel 2 during the cryogenic mission, although note that the methodology for determining pixel phase corrections was somewhat different. In particular, the cryogenic observations used more than one star placed on many different pixels across the arrays. Two-dimensional maps of the pixel phase correction will be made available in the near future.