Date: Tue, 18 Dec 2001 08:39:42 -0500 (EST) From: koekemoe@stsci.edu (Anton Koekemoer) To: dickinson@stsci.edu, donahue@stsci.edu, gak@stsci.edu, koekemoe@stsci.edu, sparks@stsci.edu Subject: Re: SHARE WCS/Astrometry Mime-Version: 1.0 Dear all, Here's the new revision of the WCS/Astrometry section - I've expanded the "goals & implementation" section and have also added a new section on resource requirements/timescales. Comments welcome - Anton. WCS/Astrometric Improvements ============================ Description: ------------ The primary purpose of this enhancement is to improve the astrometry that is encoded in the WCS keywords information in image headers, by making use of updated measurements and astrometric information about guide stars from data taken with HST. Science Case: ------------- The current astrometric accuracy of HST images is limited by the inherent astrometric uncertainties in the Guide Star Catalog system, and is generally likely to be accurate to no more than ~0.5 - 1 arcseconds. However, there is often a need to obtain much higher astrometric accuracy, in particular when comparing images at different wavebands from different telescopes (radio, optical, X-ray) as well as different images obtained with HST, generally at different times and in unrelated programs (e.g., narrow-band and broad-band images of the same object). This is required when carrying out source identifications based on multi-wavelength information, when examining color gradients across a given object, or when determining the relative location of morphological features seen in different bands. Ideally it would be desirable to achieve astrometric accuracy to a level comparable to the measurement error of unresolved sources on HST (i.e., << 0.1 arcseconds). This issue involves two separate, but related, concepts: relative astrometric precision between different HST images, and absolute astrometric accuracy with respect to some other well-defined, high-resolution system (the global VLBI reference frame is one such example). Currently, relative astrometric accuracy between different HST images is generally only achievable in cases where stars or other bright unresolved objects are common to both images. For images that have few or not stars (e.g., narrow-band or UV observations), relative astrometry can still be aimed at but becomes less certain, thereby directly impacting the science. Absolute astrometry, for example between HST images and radio data, is generally only possible in images that contain sources unresolved in both radio and optical, or otherwise display precise one-to-one correspondence in the two bands. Otherwise, if sources are resolved in one or both and display different morphologies, absolute registration becomes uncertain. Unique STScI Capability: ------------------------ STScI has the ability to update the GSC information, as well as the resources to carry out studies of all the guide stars for which astrometric information may be updated. Furthermore, only STScI has the ability to automatically incorporate the updated information into the archive data processing pipelines, thereby potentially eliminating the need for users to carry out any further refinements on the astrometry. Drawbacks: ---------- One possible drawback is that some guide stars will have better astrometric information than others, so this capability will produce improvements on a non-uniform basis for different images. Required Decisions: ------------------- * What do we use as a basis for registering images? (A) Information obtained purely from the HST images (e.g., objects in common to two or more images) and the guide stars used for each observation (B) An "external" reference frame (eg GSC-2.2, or SDSS), required to have a sufficiently high density of objects across a sufficient area of sky, combined with a sufficient degree of absolute astrometric accuracy, to be able to cross-register a usefully large number of HST images. * How do we allow users to deal with geometric distortion of HST images: - allow retrieval of images that are fully calibrated except that geometric distortion is not removed, which then calls for providing an improved version of "metric" to provide coordinates (eg the current approach for WFPC2) - Only provide "distortion-corrected" images, which would simplify astrometry between different images but would remove some degree of flexibility * What reference frame do we adopt (GSC-2/Hipparcos?) - I suspect the answer is obviously yes but it would be wise to check exactly how the current GSC-2 astrometric frame relates to Hipparcos and other systems in use at other wavebands (eg global VLBI; others?) Goals and Implementation Plans: ------------------------------- (A) Astrometry from the HST images and the guide stars used for the observation ==== * Minimum goal: Allow accurate relative astrometry between HST images from different instruments/filters that have some overlapping region of the sky in common. Implementation: at least two different options: 1) Base the registration on objects in the images themselves, either by direct cross-correlation of the overlapping regions or by means of the "engineering-grade object catalogs" if these end up being implemented. Advantage: - can be done for any images with objects in common, regardless of whether any guidestars were in common Drawbacks: - difficult to implement accurately for cases where emission-line data are involved, or even continuum data in widely separated bandpasses, due to likely spatial offsets; thus it may be limited to data from similar (or even the same!) filters - will probably not work well for images with very different exposure times - may likely be feasible for only a small fraction of images 2) Base the registration on at least one guide star in common to both images, combined with a knowledge of the roll angle offset between them. For more than two images, Advantage: - independent of the details of objects in the images Drawback: - perhaps possible for only a small fraction of images? The fraction of images that will benefit from this could in principle be exactly determined since we have the information on which guide stars are used for each image. * Medium goal: Allow accurate relative astrometry between any pair of images that have at least one guide star in common, together with knowledge of the roll angle difference between them. Implentation: Since all the available information already exists in CDBS, including information about which guidestars were used, together with the roll angle information for each exposure, this can be implemented in a relatively automated way. In practice, a query could be designed that takes a given exposure as input, and shows all the images that have a guide star in common with this input exposure (either with all instruments/filter combinations, or with some subset, as specified by the user). If the user then requests any of these subsequent exposures, their astrometric information could be recaulctaed automatically to place them in the same reference frame as the first exposure. Advantage: - independent of the details of objects in the images, and potentially allows cross-registration between many more pairs of images than in the "minimum" goal stated above. Drawbacks: - still perhaps only feasible for a relatively small fraction of the total set of HST images - if there is no overlap between images, then achieving accurate relative registration may be less interesting since less science will be enabled. However, for objects that are much more extended then the HST images, accurate relative registration may still be useful when compared to data from other wavebands that may cover the entire field (eg HST images of various separate parts of a large galaxy or extended cluster, which may have a single complete image at another waveband or from the ground). * Maximum goal: Use triangulation information from guide stars common to pairs of images, together with astrometric information from objects in the images themselves that are visible at other wavebands with high angular resolution (in particular radio), to update and improve the actual astrometric information about the guide stars. Implementation: this would be a relatively time-intensive effort, since it requires actual updating of the guide star coordindates, and relatively stringent quality control over the means by which cross-wavelength calibration is achieved (for example, many of the bright compact radio sources suitable for astrometry lie embedded in complex optical systems). This process would also be continually on-going as more HST images are obtained. A system of updating guide star coordinates would have to be designed (eg trading off the frequency of updates vs the management overhead). Advantages: - Improving the actual astrometric information of the guide stars is the only means of providing a direct tie-in to systems used at other wavelengths, most notably global VLBI reference frame. - As more HST images are obtained, paticularly with ACS (both prime parallel), the number of guide stars for which improved astrometry becomes available could dramatically increase. - This method also allows "boot-strapping": if improved astrometry is available for one guide star in a pair, the FGS locking information from the exposure can be used after-the-fact to improve the astrometry on the second guide star in the pair. Drawbacks: - Care must be taken when obtaining astrometric information from other wavebands, in ensuring that the multi-waveband emission is produced by the same physical region and is also more compact than the level of astrometric accuracy being aimed at. - Improvements in the astrometric accuracy of the guide stars will take place more or less at random and not at all uniformly. - This option is the most resource-intensive of the three, since it involves not only detailed physical understanding of the multiwavelength nature of the sources from which astrometry is being obtained, but also a substantial amount of overhead in terms of propagating the guide star position updates correctly. (B) Astrometry from an external "absolute" reference catalog. ==== Two obvious catalogs of choice are GSC-2.2 and SDSS. These are the only database that has the combination of a sufficiently large number of objects (~few per square arcminute), together with the required absolute astrometric accuracy. The description here will refer mostly to GSC-2.2, simply because it is already fully available and well-characterized while SDSS is still in progress; moreover it covers the entire sky whereas SDSS covers only 1/4. However, SDSS should prove to be a valuable addition in the area where the two catalogs overlap. * Minimum goal: Refine the astrometric keywords of an image by automatically carrying out astrometry on the known GSC-2.2 objects in the image, and calculating a correction based on the average difference between their measured coordinates and those from the GSC. Implementation: This would be reasonably straightforward to implement. For any image that is retrieved from the archive, an automatic query would be carried out that would return the identifications of any GSC objects on the image. The known position of each GSC object would be used to calculate its approximate position on the chip, and a simple centroiding algorithm would then determine the exact pixel location of the object. This would be used to determine its "measured" RA and Dec, based on the image header keywords, which would likely be offset from the catalog position. After calculating these offsets for all the GSC objects on the image, reject any obvious outliers, then determine a mean offset in RA and Dec, and apply this correction to the header keywords. Advantages: - Relatively simple to implement - Generalizable to any detector/filter combination - independent of the type of field (except that at high latitudes the GSC objects become more sparse) Drawbacks: - At high latitudes GSC objects become more sparse so this technique will likely provide non-uniform improvements depending on location in the sky - The technique will rely fairly heavily on using images cleaned of cosmic rays, otherwise too many of the GSC objects may need to be rejected. - Quantifying the "quality" of the astrometric correction may be somewhat non-trivial. * Maximum goal: This would build upon the "minimum goal", and is similar in its scope to the "maximum goal" outlined for section (A) above. Once astrometry for an image has been improved, this means that the information for the guide stars themselves should be improved as well, therefore astrometric corrections are fed back into the GSC system in an iterative process. The advantages and drawbacks are similar to case (A), and are summarized again here: Advantages: - Improving the actual astrometric information of the guide stars is the only means of providing a direct tie-in to systems used at other wavelengths, most notably global VLBI reference frame. - As more HST images are obtained, paticularly with ACS (both prime parallel), the number of guide stars for which improved astrometry becomes available could dramatically increase. - This method also allows "boot-strapping": if improved astrometry is available for one guide star in a pair, the FGS locking information from the exposure can be used after-the-fact to improve the astrometry on the second guide star in the pair. Drawbacks: - Care must be taken when obtaining astrometric information from other wavebands, in ensuring that the multi-waveband emission is produced by the same physical region and is also more compact than the level of astrometric accuracy being aimed at. - Improvements in the astrometric accuracy of the guide stars will take place more or less at random and not at all uniformly. - This option is the most resource-intensive of the three, since it involves not only detailed physical understanding of the multiwavelength nature of the sources from which astrometry is being obtained, but also a substantial amount of overhead in terms of propagating the guide star position updates correctly. Required Resources/Timescales: ------------------------------ Minimum goals: These would likely require a team consisting of a lead scientist, at least one software developer, and interaction with the pipeline team. It is also possible that a DA would be required to help with testing the products, and participate in interactions with the developer(s) and pipeline team if iterative testing is required to help identify and solve problems. Approximate resources would consist of about 0.25 FTE scientist, 0.5 FTE developer + pipeline work, 0.25 FTE DA for testing, over the course of 1 year which should be sufficient to yield the basic requirements for the minimum goals. Medium/Maximum goals: These would initially require a similar sized core team to the minimum goals, since they are predominantly extensions of the minimum goals. However, the core team would likely expand due to the need to interact with and solicit the participation of various additional staff, for example a scientist from FGS to help with guide star issues, and archive scientist(s) to assist in the implementation of corrections to the guide star coordinates. Resource estimates (wildly) are likely to run to about 0.5 - 1 FTE scientist(s), 1.5 FTE developer + pipeline / archive, 1 FTE DA(s) for testing, spread over at approximately two years.