Modelling the propagation of wildfires requires an accurate wind field to correctly predict the behaviour of the fire. Although numerical weather prediction models produce reliable and accurate mesoscale forecasts, these are typically either available at a spatial resolution many times greater than the typical resolution of a wildfire model or a spot forecast that must be spatially interpolated to the area of the modelled wildfire. Due to this, these forecasts may not account for fine-scale terrain interactions with the wind and must be downscaled to a higher spatial resolution before use in a wildfire model. These downscaling methods are typically computationally intensive, limiting their use for situations where rapid predictions are required. Despite this, a three-dimensional mass balancing method is commonly used in wildfire prediction as a preprocessing step. In this study we show that this mass balancing method can be reduced to a two-dimensional approach, greatly reducing the complexity and computational time required for the model. The two-dimensional method is compared with the existing three-dimensional method and experimentally measured datasets. Furthermore, a combination of rapid numerical solution techniques and modern computational processors allow these wind–terrain correction methods to be directly incorporated into wildfire propagation models.