Operational forecasters and weather researchers need accurate visualization of atmospheric data from both computational models and observed data. Although these two applications share some requirements, they have different needs and goals. We've developed a visualization tool for atmospheric science researchers and research weather forecasters that allows the 3D visualization of measured radar data and rendered numerical model data to show the 3D structures as well as how the weather event would look when observed in the field. Our system lets us load the original data directly onto the graphics hardware, with the grid mapping from the rendering space to the grid space programmed on the GPU. This method is flexible enough to handle the grids important in meteorological research and enables the application of advanced visualization methods available in texture-based slicing systems. The visually accurate rendering of weather data can be useful for training weather spotters, evaluating forecasting models, training forecasters to interpret radar data, and comparing sensor data to observed weather events.