Predictability of U.S. Northern Great Plains Summertime Precipitation Extremes

May-July precipitation extremes over the Northern Great Plains, comprising of Montana, North Dakota and South Dakota, produced ‘Billion Dollar Disasters’ in 1988, 1993, 2011 and 2017. The near-record low precipitation during 1988 and 2017 resulted in wildfires, livestock selloffs and agricultural losses. Near-record high precipitation during 1993 and 2011 resulted in exceptional flooding of the Missouri and Souris Rivers that destroyed property and killed five people. A better understanding of May-July precipitation predictability is key to establishing the limits of disaster early warning over the Northern Great Plains, given that approximately half of the region’s annual precipitation falls during this season. Therefore, the predictability of Northern Great Plains precipitation during May-July 1982-2017 is examined using an ensemble of retrospective forecasts and forecasts from an initialized prediction system, the North American Multi-Model Ensemble (NMME), and an ensemble of uninitialized atmospheric model simulations forced by observed time-varying boundary conditions.

The model ensembles indicate that skillful precipitation predictions over the Northern Great Plains during May-July are thwarted by the overwhelming influence of unpredictable noise relative to the predictable signal. Due to the influence of unpredictable noise, casting Northern Great Plains predictions probabilistically does not reveal a strong shift in the odds to dryness or wetness during any season. For example, while NMME predictions from the preceding April skillfully forecast precipitation during the extreme May-July seasons of 1988 and 1993, the probability of recording an outcome in the lowest and highest terciles, respectively, during those seasons was only around 50%. By contrast, the May-July precipitation extremes during 2011 and 2017 were not skillfully forecast by NMME at any lead time.

The model simulations suggest that El Nino Southern Oscillation (ENSO) is one source of the limited predictability of May-July Northern Great Plains precipitation. El Nino is related to a slight shift in the odds to wetness while La Nina is related to a slight shift in the odds to dryness over the region. This is the time of year when the state of ENSO is least predictable in advance.