Most global water cycling occurs over the ocean, with ~80% of Earth’s surface water fluxes (evaporation/precipitation) and 97% of Earth’s free water by storage, residing there. The 10% difference between oceanic evaporation (source) and precipitation (sink) is Earth’s primary terrestrial water source. This water is transported over land, eventually falling as terrestrial precipitation. As oceanic moisture evaporates, it leaves a sea surface salinity (SSS) signature, providing the link that enables ocean salinity to predict terrestrial precipitation. Previous work has shown salinity to be a useful predictor for summertime U.S. precipitation one season ahead. However, this study is the first in published literature to assess the role of SSS on subseasonal timescales, particularly on low-skill summertime precipitation predictions. We approach this problem using a neural network using CESM2 Large Ensemble model data to quantify the predictability of the U.S. Midwest summertime heavy rainfall events for zero to eight-week leads guided by time-varying Atlantic SSS anomalies. Neural network output allows us to identify predictions that result in forecasts of opportunity, e.g., predictions that are both confident and accurate. Using explainable artificial intelligence, we create heatmaps of the most sensitive regions of SSS anomalies in the tropical and North Atlantic, providing skillful forecasts of opportunity. To further evaluate linkages, a moisture tracking algorithm is applied to reanalysis, determining evaporation sources that are transported and fall as Midwest precipitation. These results complement the explainable artificial intelligence findings to reveal robust and physically meaningful sources of summertime Midwest precipitation predictability via Atlantic SSS anomalies. Such new insights allow early warning system augmentation to predict future heavy precipitation events, yielding enhanced lead times and improved skill.