Automated Disaggregation of Data from the National Energy Modeling System

Publication Date


Committee Chair

Karen K. Kemp, Ph.D.

Committee Members

Ray Carnes


The National Energy Modeling System (NEMS) is an all-sector, integrated model of the U.S. energy system. NEMS is used to produce the official annual 25 year forecast of U.S. energy use aggregated to the nine-region Census Division level. The objective of this project is to reduce (disaggregate) the aggregated regional energy forecast to the county level for use in a more detailed and accurate analysis of energy usage across the US.

The process of disaggregation was researched and a model was created utilizing available population forecasts and climate zone data. The model’s primary purpose was to generate an energy demand forecast that is more accurate than the commonly used models based on predicted population growth. The methods developed were then applied to the study data to obtain residential and commercial electricity demand forecasts. The model was then subjected to comparative and statistical testing to assess predictive accuracy. Forecasts using this model were robust and accurate in slow-growing, temperate regions such as the mid-west and mountain regions. Interestingly, however, the model performed with less accuracy in the pacific and northwest regions of the country where population growth was more active. In the future more refined methodologies will be necessary to increase the accuracy of these forecasts.

The disaggregation method was written into a flexible tool within the ArcGIS environment which enables the user to output the results, in five year intervals, over the period 2000-2025. In addition, the outputs of this tool were used to develop a time-series simulation showing the temporal changes in electricity forecasts in terms of absolute, per capita and density of demand.

Full text is available at the University of Redlands