Abstract
Heating and cooling degree-days (HDDs and CDDs) link outdoor temperature to building heating and cooling demand and serve as inputs for building energy and emissions models. Existing data-driven forecasts cover a few stations or provinces within single countries; none exists at national ZIP-code resolution for the United States. This study forecasts monthly HDDs and CDDs from 2026 to 2040 for 2903 ZIP codes spanning the climatic range of the United States, using 2000 to 2025 records from the DegreeDays.net API at a base temperature of 65 °F (18.3 °C). Each series is modeled with a trend-augmented seasonal ARIMA model and a damped-trend exponential smoothing model, combined in a geometric-mean ensemble and benchmarked against a seasonal-naive baseline, with Monte Carlo prediction intervals. In rolling-origin backtesting, the ensemble and damped-trend models achieve a mean MASE between 0.84 and 0.93 over the fifteen-year horizon. Across the national portfolio, HDDs are projected to decline by 9.01% and CDDs to rise by 25.99% between 2025 and 2040; HDDs fall in 94.7% of ZIP codes and CDDs rise in 96.2%. National forecast trends match LOCA2 downscaled CMIP6 projections under SSP2-4.5. The forecasts provide a reproducible, ZIP-code-resolution basis for municipal energy planning and building performance compliance.