This brief article describes refinements to the Graves and Pitarka (2010) broadband ground motion simulation methodology (GP2010 hereafter) that have been implemented in version 14.3 of the SCEC Broadband Platform (BBP). The updated version of our method on the current SCEC BBP is referred to as GP14.3. Our simulation technique is a hybrid approach that combines low-­‐frequency and high-­‐frequency motions computed with different methods into a single broadband response. The separate low-­‐ and high-­‐frequency components have traditionally been called “deterministic” and “stochastic”, respectively; however, this nomenclature is an oversimplification. In reality, the low-­‐frequency approach includes many stochastic elements, and likewise, the high-­‐frequency approach includes many deterministic elements (e.g., Pulido and Kubo, 2004; Hartzell et al., 2005; Liu et al., 2006; Frankel, 2009; Graves and Pitarka, 2010; Mai et al., 2010). While the traditional terminology will likely remain in use by the broader modeling community, in this paper we will refer to these using the generic terminology “low-­‐frequency” and “high-­‐ frequency” approaches. Furthermore, one of the primary goals in refining our methodology is to provide a smoother and more consistent transition between the low-­‐ and high-­‐ frequency calculations, with the ultimate objective being the development of a single unified modeling approach that can be applied over a broad frequency band. GP2010 was validated by modeling recorded strong motions from four California earthquakes. While the method performed well overall, several issues were identified including the tendency to over-­‐predict the level of longer period (2-­‐5 sec) motions and the effects of rupture directivity. The refinements incorporated in GP14.3 are aimed at addressing these issues with application to the simulation of earthquakes in Western US (WUS). These refinements include the addition of a deep weak zone (details in following section) to the rupture characterization and allowing perturbations in the correlation of rise time and rupture speed with the specified slip distribution. Additionally, we have extended the parameterization of GP14.3 so that it is also applicable for simulating Eastern North America (ENA) earthquakes. This work has been guided by the comprehensive set of validation studies described in Goulet and Abrahamson (2014) and Dreger et al. (2014). The GP14.3 method shows improved performance relative to GP2010, and we direct the interested reader to Dreger et al. (2014) for a detailed assessment of the current methodology. In this paper, we concentrate on describing the modifications in more detail, and also discussing additional refinements that are currently being developed.