California Winter Precipitation Predictability: Insights From the Anomalous 2015-2016 and 2016-2017 Seasons

The unexpected dry 2015-2016 El Nino winter and extremely wet 2016-2017 La Nina winter in California challenged current seasonal prediction systems. Using the Met Office GloSea5 forecast ensemble, we study the precipitation and circulation differences between these seasons and identify processes relevant to California precipitation predictions. The ensemble mean accurately predicts the midlatitude atmospheric circulation differences between these years, indicating that these differences were predictable responses to the strong oceanic forcing differences. The substantial California precipitation differences were poorly predicted with large uncertainty. Notable differences in high-latitude circulation anomalies associated with internal variability distinguish the ensemble members that successfully simulate precipitation from those that do not. Specifically, accurate representation of the Arctic Oscillation phase differences improves the accuracy of simulated precipitation differences but these differences were not well predicted in the ensemble mean for these seasons. Improved representation of high-latitude processes such as the Arctic Oscillation and polar-midlatitude teleconnections could therefore improve California seasonal predictions. Plain Language Summary California recently experienced two unusual winter seasons. Following a failed rainy season despite the strong 2015-2016 El Nino that typically brings heavy rains, California unexpectedly experienced one of its wettest winter seasons on record during the 2016-2017 La Nina. The seasonal forecast systems were unable to predict these unusual winter precipitation patterns. We examine the ability of a state-of-the-art forecast system to capture the large-scale atmospheric circulation patterns that influence the track of storms, which bring a majority of winter precipitation to California. We show that the seasonal forecast systems can reproduce the large-scale circulation differences in the North Pacific-American domain, but random atmospheric variability can still easily prevent the accurate prediction of regional scale precipitation in extratropical regions such as California. Further, we identify the role of a natural pattern of large-scale variability in the atmosphere that affects weather in the middle and high latitudes, referred to as the Arctic Oscillation, in controlling the accuracy of California precipitation forecasts. Prioritizing improvements in the representation of these patterns, the processes by which they are predicted and their influence over other regions in the forecasts systems can help improve seasonal predictions, which is important for the management of California's water resources and infrastructure.