Quantifying the relationship between temperature fluctuations and energy usage can be useful for many aspects of planning for ongoing operations and future economic development. When combined with projections of 21st century climate change it is possible to develop new insights into how energy demand may be influenced by climate decades into the future.

This paper describes the development of an objective technique to relate the monthly and seasonal residential consumption of natural gas in the U.S. to atmospheric temperature and its fluctuations, focusing on the cold part of the year when residential demand is dominated by heating. This technique improves on previous work through the use of a new analysis technique to better gauge the impact of daily weather extremes. The objective is to quantify the relationship between daily temperature fluctuations and monthly residential natural gas demand with the specific goal of creating a statistical model that estimates how anomalous temperatures affect natural gas demand on monthly and seasonal timescales. The model indicates the percent departure of natural gas demand from the long-term average given accurate temperature variations. The model could be used with forecast temperatures to predict future natural gas departures. The approach is based on calculation of days below a specified temperature threshold that is different for each location, and thus differs from conventional “heating degree day” approaches. The threshold for each station is determined as a percentile boundary in the daily temperature distribution observed over a defined period (1987-2004). A national index for a given month or group of months is then constructed by weighting each station temperature using the population and the estimated long-term mean gas consumption of the surrounding region.

This paper also addresses the extent to which natural gas demand will change in the future based on long-term projections of the atmospheric warming expected to result from anthropogenic forcing. Specifically the results of a forecast of temperature in the late 21st century versus late 20th century made by the MIROC3.2 “high-resolution” global climate model (Hasumi and Emori, 2004) is used together with our statistical gas demand model. Based on these forecast temperatures, the results suggest that, holding non-climate-related factors constant, the U.S. cold season residential natural gas consumption can be expected to decline substantially over the next century.