Using a statistical model to verify warm conveyor belts in ECMWF’s sub-seasonal forecasts

Rapidly ascending air streams in midlatitude cyclones – so-called warm conveyor belts (WCBs) – affect the lifecycle of blocking anticyclones. WCBs are usually identified by selecting coherent bundles of rapidly ascending trajectories. Their calculation, however, requires data at a high spatio-temporal resolution and is computationally expensive. To identify WCBs in expansive data sets such as ensemble reforecasts or climate model projections, alternative approaches are necessary. 

In this study we introduce a logistic regression model which is capable of identifying the inflow, ascent, and outflow phase of WCBs based on Eulerian input parameters. Validation against a Lagrangian-based dataset confirms that the logistic model is reliable in replicating the climatological frequency of WCBs as well as the footprints of WCBs at instantaneous time steps. 

Second, we employ the statistical model to verify the representation of WCBs in ECMWF’s sub-seasonal reforecasts. Overall the reforecasts depict frequencies of WCBs across seasons relatively well at all lead times. A correction of biases in the meteorological parameters for the logistic model partly removes existing biases in the reforecast WCB climatology. However, the bias-corrected  forecast skill still rapidly decays leaving useful skill only up to around day 8. These results corroborate that synoptic-scale activity might hinder accurate forecasts into sub-seasonal time scales for the extratropical large-scale circulation. Future work will elucidate if and in which situation poor skill for WCBs dilutes skill for Atlantic-European weather regimes on sub-seasonal time scales.