On the configuration of a regional Arctic Numerical Weather Prediction system to maximize predictive capacity

Abstract

Limitations to operational weather forecasts exist in terms of availability of computer (and human) resources combined with operational deadlines. For operational weather services it is therefore important to utilize their resources to maximize the predictive capability. This study shows how forecast quality in a state-of-theart high-resolution regional Arctic Numerical Weather Prediction (NWP) system changes with varying configuration choices; (1) Ensemble Prediction System (EPS), (2) higher spatial resolution, (3) atmospheric initialization by assimilation of observations, (4) surface initialization by assimilation of observations and by (5) changing the regional domain and location. Results from such inter-comparisons are useful guidance for (Arctic) weather forecast systems, and can together with information on e.g. user-needs and post-processing capabilities be used to maximize the operational predictive capacity. All configuration choices have a significant impact on the forecast quality of near-surface parameters, but the impact varies with parameter, region, weather type, lead time and part of the forecast evaluated (e.g. average errors or rare events). Higher spatial resolution and EPS are expensive, but are still promising to further improve state-of-the-art regional Arctic high-resolution NWP systems. In particular when forecasting rare events regional EPS shows huge benefits. Assimilation of observations in the initialization process of the regional NWP system has also a positive impact on forecast quality. Finally, although less pronounced, the choice of the domain size and location also has a significant impact and should therefore be chosen carefully.

IWIN: the Isfjorden Weather Information Network

“…Furthermore, observations from IWIN address the need for improving our weather and climate prediction capabilities in the Arctic, which are not as good as at lower latitudes (Køltzow et al, 2021). IWIN does so by providing additional observations from an otherwise data-sparse region, both for better process-level understanding (and thereby model and parameterization development) and for data assimilation pur- poses.…”