Comments on “Characterizing ENSO Coupled Variability and Its Impact on North American Seasonal Precipitation and Temperature”

El Niño and La Niña seasonal weather anomaly associations provide a useful basis for winter forecasting over the North American regions where they are sufficiently strong in amplitude and consistent in character from one event to another. When the associations during La Niña are different than El Niño, however, the obvious quasi-linear-statistical approach to modeling them has serious shortcomings. The linear approach of L’Heureux et al. is critiqued here based on observed land surface temperature and tropospheric circulation associations over North America. The La Niña associations are quite different in pattern from their El Niño counterparts. The El Niño associations dominate the statistics. This causes the linear approach to produce results that are inconsistent with the observed La Niña–averaged associations. Further, nearly all the useful North American associations have been contributed by the subset of El Niño and La Niña years that are identifiable by an outgoing longwave radiation (OLR) El Niño index and a distinct OLR La Niña index. The remaining “non-OLR events” exhibit winter weather anomalies with large event-to-event variability and contribute very little statistical utility to the composites. The result is that the linear analysis framework is sufficiently unable to fit the observations as to question its utility for studying La Niña and El Niño seasonal temperature and atmospheric circulation relationships. An OLR-event based approach that treats La Niña and El Niño separately is significantly more consistent with, and offers an improved statistical model for, the observed relationships.