Skillful long-lead prediction of summertime heavy rainfall in the US Midwest from sea surface salinity

Abstract

Summertime heavy rainfall and its resultant floods are among the most harmful natural hazards in the US Midwest, one of the world's primary crop production areas. However, seasonal forecasts of heavy rain, currently based on preseason sea surface temperature anomalies (SSTAs), remain unsatisfactory. Here, we present evidence that sea surface salinity anomalies (SSSAs) over the tropical western Pacific and subtropical North Atlantic are skillful predictors of summer time heavy rainfall one season ahead. A one standard deviation change in tropical western Pacific SSSA is associated with a 1.8 mm day−1 increase in local precipitation, which excites a teleconnection pattern to extratropical North Pacific. Via extratropical air-sea interaction and long memory of midlatitude SSTA, a wave train favorable for US Midwest heavy rain is induced. Combined with soil moisture feedbacks bridging the springtime North Atlantic salinity, the SSSA-based statistical prediction model improves Midwest heavy rainfall forecasts by 92%, complementing existing SSTA-based frameworks.