Near-future astronomical survey experiments, such as LSST, possess system requirements of unprecedented fidelity that span photometry, astrometry and shape transfer. Some of these requirements flow directly to the array of science imaging sensors at the focal plane. Availability of high quality characterization data acquired in the course of our sensor development program has given us an opportunity to develop and test a framework for simulation and modeling that is based on a limited set of physical and geometric effects. In this paper we describe those models, provide quantitative comparisons between data and modeled response, and extrapolate the response model to predict imaging array response to astronomical exposure. The emergent picture departs from the notion of a fixed, rectilinear grid that maps photo-conversions to the potential well of the channel. In place of that, we have a situation where structures from device fabrication, local silicon bulk resistivity variations and photo-converted carrier patterns still accumulating at the channel, together influence and distort positions within the photosensitive volume that map to pixel boundaries. Strategies for efficient extraction of modeling parameters from routinely acquired characterization data are described. Methods for high fidelity illumination/image distribution parameter retrieval, in the presence of such distortions, are also discussed.