Academic perspective: Forecasting temperature (and HDDs and CDDs) with lead times of hours to decades

The goal of the NEHRT Program has been to improve the accuracy and localization of forecasts of daily maximum and minimum surface temperature in New England, with a societal goal of reducing costs and conserving energy. Universities can contribute to this goal by providing student research opportunities focusing site-specific forecast accuracy with lead times of importance to the utility industry. Two examples with quite different lead times will be described. Changnon, et al., 1995 concluded from a survey of utility companies that tasks relating to power trading, fuel acquisition, load forecasting, and systems planning made frequent use of weather and climate forecasts because such forecasts were judged to be of high value (>$100,000) to the company. From this we can conclude that accurate weather and climate information, temperature and heating/cooling degree days in particular, are needed by utilities on short lead times for matching supply with demand and arranging for near-term future supplies as well as long lead times for rate-setting with state utility boards and infrastructure planning. Forecasting bridge frost will be described as an example of a localized short-term temperature forecast that demonstrates forecast accuracy under conditions when variable cloudiness is not a factor, and also the degree to which temperature forecasts can be degraded by clouds. Long-term utility planning requires information relating future trends and variability from “normal”. Examples will be shown that recent observed trends and projected future changes in climate underscore the need for revised definitions of what is “normal” and how utilities should use revised statistical methods as well as regional climate change.