The Wang-Sheeley-Arge (WSA) model estimates the solar wind speed and interplanetary magnetic field polarity at any point in the inner heliosphere using global photospheric magnetic field maps as input. WSA employs the Potential Field Source Surface (PFSS) and Schatten Current Sheet (SCS) models to determine the Sun’s global coronal magnetic field configuration. The PFSS and SCS models are connected through two radial parameters, the source surface and interface radii, which specify the overlap region between the inner SCS and outer PFSS model. Though both radii are adjustable within the WSA model, they have typically been fixed to 2.5 R sol. Our work highlights how the solar wind predictions improve when the radii are allowed to vary over time. Data assimilation using particle filtering (sequential Monte Carlo) is used to infer the optimal values over a fixed time window. The Air Force Data Assimilative Photospheric Flux Transport (ADAPT) model generates an ensemble of photospheric maps that are used to drive WSA. When the solar wind model predictions and satellite observations are used in a newly-developed quality-of- agreement metric, sets of metric values are generated. These metric values are assumed to roughly correspond to the probability of the two key model radii. The highest metric value implies the optimal radii. Data assimilation entails additional choices relating to input realization and timeframe, with implications for variation in the solar wind over time. We present this work in its theoretical context and with practical applications for prediction accuracy.