Dias, D. F., D. R. Cayan, A. Gershunov and A. J. Miller, 2022:

The influence of sea surface temperature and soil moisture in seasonal predictions of air temperature over Western North America

Journal of Climate, sub judice.

Abstract. Seasonal predictions have the potential to improve the management of different sectors of the society by anticipating climate fluctuations and possible weather extremes. Such forecasts must contend with a high level of natural variability as well as challenges posed by climate change. However, they are constrained by limited understanding of local and regional atmospheric predictability. Here, a canonical correlation analysis (CCA) prediction model of minimum and maximum air temperature anomalies (Tmin and Tmax) over Western North America (WNA) is developed. Remote and local predictors are used: sea surface temperature (SST) across the Pacific and local soil moisture (SM). The evaluation of the skill of predicted air temperature using historical observations indicates that CCA can provide skillful predictions for seasonal anomalies of air temperature over the region. However, skill is found to vary over seasons, location and combination of predictor and predictand variables. SST yields the best predictive skill for Tmax and Tmin during wintertime, but for spring and early-summer its influence is mostly on Tmin. Remote large-scale patterns, in the form of climate indices, are captured by the CCA canonical modes and it is shown that they can be responsible for this predictive ability. On the other hand, the influence of SM is restricted to Tmax and only during the winter, when it is shown that SM has the highest autocorrelation for the region. The results demonstrate the importance of careful analyses that consider season, variable being predicted, and predictors in forming statistical forecast models to be used for decision making.

Preprint (pdf)