Rapid 30-Second-Update Numerical Nowcasting with Multi-Parameter Phased Array Radar Observations: Skill Comparison with an Advection Model

Abstract Precipitation nowcasting consists of high-resolution forecasting of rainfall at very-short lead times and is crucial for assessing the imminent risk from heavy rainfall, often from fast-evolving convective weather systems that can develop rapidly within minutes. Common approaches to precipitation nowcasting include the use of a convection-resolving numerical weather prediction (NWP) model or advection-based model, the latter of which rely on temporal extrapolations of rainfall based upon flow vector calculations from quantitative precipitation estimation images. With the development of the Multi-Parameter Phased Array Weather Radar (MP-PAWR) providing high-resolution observations every 30-seconds, the potential to refresh these models at frequencies high enough to capture the rapid evolution of convective weather systems is possible. In this study we compare nowcasting skill using a regional-scale NWP model and an advection model that both employ a 30-second update using MP-PAWR observations for a single convective event that brought heavy rainfall to Tokyo in 2021. Based on a skill evaluation of 120 forecasts, this study highlights the advantage of the NWP system over the advection-based model to better predict rainfall from 2 to 30 min, led by its ability to predict rapid physical changes to the convective system over this timeframe.