Identification of Moisture Transport Pathways Using a Self-Organizing Map (SOM) Method

This study employs a self-organizing map (SOM) method to identify pathways of moisture transport in the northeastern United States. Input data for the SOM are integrated vapor transport (IVT) calculated from daily ERA-Interim reanalysis eastward and northward vertically integrated vapor fluxes at 1200 UTC from 1985 through 2015 at a spatial resolution of 0.75° × 0.75°. The study region was bounded by 36°N to 51°N and 85°W to 60°W, centered on the northeastern United States. The SOM was trained over these data in 2000 iterations. The optimal number of patterns retained was determined by minimizing the variance of days within each pattern while maximizing the variance between patterns. The resulting maps, each displaying a unique pattern of water vapor transport, represent the pathways by which moisture moves within the study region. While some maps indicate very little water vapor flux, others show narrow bands of enhanced vapor transport, which may be indicative of atmospheric rivers influencing the region. Identification of these patterns and pathways of moisture transport is important in the northeastern United States as the region is characterized by agriculture as well as large urban areas, which are both dependent on water resources. As such the results of this study may be applied to water resource management practices as well as used to provide a baseline should this study be replicated using model output of future climate in the context of global change.