An atmospheric dynamics perspective on the amplification and propagation of forecast error in numerical weather prediction models: A case study. (12th November 2018)

• Record Type: Journal Article
• Authors: Grams, Christian M.; Magnusson, Linus; Madonna, Erica
• Is Part Of: Quarterly journal of the Royal Meteorological Society. Volume 144:Number 717(2018)
• Page Start: 2577
• Page End: 2591
• Publication Date: 2018-11-12
• Subjects: forecast bust -- numerical weather prediction -- potential vorticity -- predictability -- upscale error growth -- warm conveyor belt -- weather regimes; Meteorology -- Periodicals; 551.5
• DOI: 10.1002/qj.3353
• Languages: English
• ISSNs: 0035-9009

Abstract

Abstract : Despite huge progress made, state‐of‐the‐art numerical weather prediction systems occasionally experience severe forecast busts for the large‐scale extratropical circulation. This study investigates one of the most severe forecast busts for Europe in the European Centre for Medium‐Range Weather Forecasts integrated forecasting system (IFS) in recent years. The forecast bust occurred in March 2016 and was associated with a misforecast of the onset of a blocking regime. We investigate the evolution of the forecast error in the IFS ensemble by employing a potential vorticity perspective combined with Lagrangian diagnostics. We show that the error grows rapidly from an initially small perturbation in the detailed structure of an upper‐level trough near Newfoundland. This trough triggers strong diabatic warm conveyor belt activity in the North Atlantic region. The misrepresentation of this warm conveyor belt activity in the ensemble forecast amplifies the initial condition error and communicates it downstream into Europe. Specifically, the ensemble underestimates poleward warm conveyor belt ascent and associated warm conveyor belt outflow into high latitudes. Instead, all ensemble members forecast too strong warm conveyor belt outflow further to the south, which ultimately results in a wrong forecast of the upper‐level Rossby wave pattern over Europe. This case study shows that warm conveyor belts and the associated latent heat release in slantwise ascending air can trigger a nonlinear feedback mechanism that amplifies forecast error strongly and communicates it into regions far downstream. It corroborates the fact that multiscale interactions and moist‐and dry‐dynamical processes ranging from microphysical to synoptic scales need to be represented accurately in numerical weather prediction, in order to predict the extratropical large‐scale circulation correctly. Abstract : The onset, decay, and transition of large‐scale flow regimes constitute a key challenge in subseasonal numerical weather prediction and are prone to forecast error. In this study, we document one of the severest forecast busts in the ECMWF IFS ensemble in recent years. We show that, during the onset of European blocking, a poorly forecast warm conveyor belt amplified and propagated downstream an initial condition error, corroborating the need to represent cloud diabatic processes correctly in order to improve extended‐range forecasts.