How warm conveyor belt activity across the North Pacific influenced the predictability of the North American heat wave 2021

In June 2021, the western North American continent experienced an intense heat wave with unprecedented temperatures and far-reaching socio-economic consequences. The magnitude of the heat wave was substantially underestimated by probabilistic weather forecasts for lead times beyond seven days. The record-breaking temperature anomaly coincided with a far northward extending upper-level ridge that was unambiguously linked to the intensity of the heat wave. During the 10 days preceeding the heat wave, the upper-level ridge was continuously fed by air masses originating to a substantial fraction from the lower troposphere that ascended in the West, Central, and East Pacific.
We analyze the role of strongly ascending airstreams, so called warm conveyor belts (WCBs), for the upper-level ridge amplitude, and illustrate how the anomalous WCB activity in the North Pacific limits the predictability horizon of this extreme event. We identify footprints of WCBs in operational ensemble forecasts from the European Centre for Medium-Range Weather Forecasts which is enabled through a machine-learning based diagnostic. The 51 member ensemble with lead times up to 15 days is stratified into a subset that best capture the upper-level ridge and potential vorticity anomaly ('good' members), and one with the largest discrepancy in the upper-level flow field ('bad' members). We find that the underestimation of the ridge amplitude over the North American continent in the bad forecasts is associated with a mis-representation of WCB activity across the West and East Pacific. While WCB outflow in the East Pacific maintains the ridge quasi-stationary and re-amplifies the pre-existing PV anomaly, WCB outflow in the West Pacific lifts the tropopause to anomalous heights and strengthens the upper-level jet, which facilitates East Pacific WCB ascent through downstream development. The mis-representation of this chain of synoptic events in the bad members finally results in an erroneous position and amplitude of the upper-level ridge and associated temperature anomaly. We conclude that the chain of synoptic events across the Pacific and their model representation play an essential role for the upper-level ridge position and amplitude and limit the predictability of the magnitude of the heat wave.