Spring warmups and snowmelt in the western United States: Predictability on medium-range timescales with CDC reforecast ensembles

In the western United States, the timing of spring snowmelt and associated flood peaks (flushing flows) have important implications for reservoir management and riparian ecosystems. Skillful medium-range temperature and circulation forecasts can augment the lead-times for hydrologic prediction and improve early decision-making.

We pursue a retrospective assessment of the spring snowmelt periods for the years 1979-2004 to understand the scales and characteristics of antecedent warmup episodes, with a focus on the predictability of associated large-scale atmospheric precursors. Atmospheric signatures common to spring warmups are isolated and historical CDC forecast ensembles are examined to quantify the models' ability to forecast these episodes at up to two-week leads. The historical CDC reforecast ensembles facilitate the use of analog forecasting techniques to correct the ensemble forecasts. Utilizing atmospheric indices tailored to this problem, a simple methodology is proposed to generate analog-based forecasts. Initially, this methodology will be applied to large-scale warmups over the western U.S. but it also could be applied at the basin scale in the future. Additional factors influencing the timing and magnitude of streamflow peaks, such as those related to snowpack characteristics, are also discussed.