Exploring the Efficacy of Multivariate Air Masses as a Predictor of Electricity Demand: A Case Study in the Northeastern United States

ABSTRACT In many midlatitude regions, human thermal comfort and electricity demand are strongly linked via the use of heating and air conditioning. Human biometeorological research has shown that the relationship between humans and their thermal comfort is quite complex, and a multitude of different thermal comfort metrics have been developed to examine it. Based upon prior research that has shown air masses (AMs) influence human thermal comfort, herein we examine whether AMs from version 2 of the gridded weather typing classification (GWTC‐2) can be used to model electricity demand anomalies in the northeastern United States. Results show that AMs are indeed significant predictors of demand, especially in the summer and winter. In summer, concurrently humid and warm conditions demand up to 77 GWh/day more electricity and are associated with a near 13‐fold increase in the risk of an electricity demand spike in some months, whereas a dry‐warm AM does not show any significant results. In some winter months, electricity demand rises by 39 GWh/day when a dry‐cool AM occurs, with an 18× risk of an electricity demand spike. When used in modelling, the AMs are generally slightly better predictors than dry‐bulb temperature, and model electricity demand with minimal bias, small errors, and explain about 94% of the variability in day‐to‐day electricity demand. Although this study is limited in scope, with this proof‐of‐concept established, future research should expand the examination of AMs in electricity demand modeling to include more geographic regions, other pertinent industry outcomes (e.g., capacity), and more advanced modeling techniques.