Designing Imaging Surveys for a Retrospective Relative Photometric Calibration

In this article, we investigate the impact of survey strategy on the performance of self-calibration when the goal is to produce accurate photometric catalogs from wide-field imaging surveys. This self-calibration technique utilizes multiple measurements of sources at different focal-plane positions to constrain instruments' large-scale response (flat-field) from survey science data alone. We create an artificial sky of sources and synthetically observe it under four basic survey strategies, creating an end-to-end simulation of an imaging survey for each. These catalog-level simulations include realistic measurement uncertainties and a complex focal-plane dependence of the instrument response. In the self-calibration step, we simultaneously fit for all the star fluxes and the parameters of a position-dependent flat-field. For realism, we deliberately fit with a wrong noise model and a flat-field functional basis that does not include the model that generated the synthetic data. We demonstrate that with a favorable survey strategy, a complex instrument response can be precisely self-calibrated. We show that returning the same sources to very different focal-plane positions is the key property of any survey strategy designed for accurate retrospective calibration of this type. The results of this work suggest the following advice for those considering the design of large-scale imaging surveys: Do not use a regular, repeated tiling of the sky; instead, return the same sources to very different focal-plane positions.