A New Framework for Understanding Systematic Errors in Cluster Lens Modeling. IV. Constraints from the Giant Arc in Abell 370

Abstract We aim to improve cluster lens modeling and source reconstruction by utilizing the full information in giant, caustic-crossing arcs lensed by galaxy clusters. Lens models are generally constrained using image positions and assuming point sources, but spatially extended giant arcs provide more constraints; however, they require a more complex model that accounts for the structure of the extended source. We seek to determine whether improvements to the lens model and reconstructed source merit the difficulty of handling the extra constraints. We choose the spatially extended z = 0.725 giant arc in the z = 0.375 Abell 370 galaxy cluster field for our study. We present (1) a series of pixel-based source reconstructions (PBSRs) for cluster mass models exploring the range of uncertainties in our fiducial model, (2) a similar analysis done using a prototype Python de-lensing code for cluster mass models from each of the Hubble Frontier Fields (HFF) modeling teams, (3) an optimized model with PBSR, and (4) and an investigation of how our optimized model affects the cluster mass model locally and globally in the highest-magnification regions. We find that our optimized model (1) is able to correct resolution-limited assumptions in cluster model inputs local to the arc, (2) has significantly smaller arc model residuals than results from the standard HFF models, and (3) affects the critical curves and therefore the information derived from highest-magnification zones most significantly in regions local to the arc.