An Inverse Modelling Framework for SAR Imagery Analysis Built on SARCASTIC v2.0

The analysis and interpretation of SAR imagery is widely recognised to be a challenging problem. The number and nature of non-intuitive effects often complicate human visual analysis, whilst the wide variation of target scattering behaviour over extended operating conditions is well-known to hinder automated processing. In this paper, we demonstrate how the SARCASTIC simulation engine is being extended to support answering analytical questions which would typically require significant input from expert analysts through inverse modelling approaches. This is a critical step towards enabling the exploitation of SAR collections at scale, which will become increasingly important as New Space continues to increase the number of taskable sensors on orbit. A case study examining an unusual urban target is presented, utilising real SAR collection data from an Umbra SAR satellite and matching simulations produced using the SARCASTIC toolset. An example analytical approach is demonstrated, and several opportunities for model-aided exploitation are identified. Preliminary results demonstrate promising correlation between the real and synthetic datasets.