Validation of precipitation forecasts from the AROME numerical weather prediction model in Norway
Master thesis
View/ Open
Date
2022-11-21Metadata
Show full item recordCollections
- Master theses [125]
Abstract
High-quality precipitation forecasts are key to ensure the public and economic safety during severe precipitation events, and the process of validating these forecasts is a continuous effort. But with Norway's varied climate and landscape, this can pose a challenge. In this thesis, the precipitation forecast data from the post-processed AROME-MetCoOp model were validated against observational data over a period from 1. December 2019 - 31. April 2022 in these six locations: Bergen, Oslo, Trondheim, Tromsø, Kristiansand and Nesbyen. The results were split into a climatology part and verification part. For climatology, Bergen and Tromsø forecasted way too little total precipitation, with the biggest deviation during summer for Bergen and winter for Tromsø. This was not due to a bias on mean precipitation amount in the model, but it could be due to the model underestimating the orographic enhancement. The model predicted a bit too much winter precipitation Oslo, Kristiansand and Nesbyen, which could be related to wind-induced undercatch of solid precipitation, although more research is needed. Precipitation distribution seemed to be somewhat narrow overall, forecasting too many low-intensity precipitation events, but struggling to forecast enough extreme precipitation. For verification results, forecast quality remained fairly constant with increasing forecast lengths (up to +48h ahead), and improved slightly with longer accumulation lengths (also up to 48h). It looks like the model performs better when the (high) hourly variability gets averaged out. All in all, Kristiansand was the best-performing location, while Tromsø saw the poorest results.