Forecast Errors and Uncertainties in Atmospheric Rivers

Observational campaigns use a range of instruments to probe atmospheric and oceanic processes to improve their understanding, which can then aid the development of numerical weather prediction. In winter 2018 and 2019, the Atmospheric River Reconnaissance (AR Recon) campaign released dropsondes and radiosondes into atmospheric rivers (ARs) over the Northeast Pacific Ocean in part to collect unique observations of ARs. These synoptic features can be responsible for extreme precipitation and flooding in coastal mountainous regions and can influence the atmospheric dynamics and predictability.

In this research, the AR Recon dropsonde observations are used to evaluate the forecasts from the European Centre for Medium-Range Weather Forecasts Integrated Forecasting System (IFS). Results show that the short-range forecasts in the IFS data assimilation system have (1) a cold bias of up to 0.7K throughout the atmosphere and (2) a dry bias and weaker winds at low altitudes (>850hPa) which causes weaker IFS water vapor fluxes. Also, in the medium-range horizon, the water vapor flux forecasts are found to be underdispersive. These errors will affect the model’s hydrological cycle and precipitation forecasts. Future research may consider using the radiosonde network to determine if these errors exist across all atmospheric conditions.