The compressible nature of the supersonic combustion flow field results in significant temperature and pressure fluctuations. In the standard Flamelet/Progress Variable (FPV) model, not only is it difficult to specify the tabulation boundary conditions for the flamelet database, but using a single fixed flamelet database to describe the entire complex flow field will likely introduce significant errors. To account for this, an improved FPV model is proposed in this study. Dynamic adjustment of the tabulation pressure in the flamelet database is achieved by scaling the benchmark flamelet database to match the local pressure. Real-time calibration of the tabulation oxidant temperature and fuel temperature is also achieved by real-time statistics and interpolation between four flamelet databases. This allows the improved FPV model to adaptively match the target supersonic combustion flow field without the difficulty of specifying tabulation pressures and temperatures as in the standard FPV model, and it avoids the limitations imposed by the single flamelet database. An additional species transport equation is also incorporated into this model to improve species prediction. Based on the simulations of a hydrogen-fueled scramjet with low-and-high total temperature, the improved model is shown to better restore wall-pressure profiles, combustion mode predictions and OH (hydroxyl radical) distributions, compared to the standard FPV model and comparable studies. This indicates that the effects of compressibility on the chemical reaction rates have been considered.