Weather Satellites and the Economic Value of Forecasts: Evidence from the Electric Power Industry

Data from weather satellites have become integral to the weather forecast process in the United States and abroad. Satellite data are used to derive improved forecasts for short-term routine weather, long-term climate change, and for predicting natural disasters. The resulting forecasts have saved lives, reduced weather-related economic losses, and improved the quality of life. Weather information routinely assists in managing resources more efficiently and reducing industrial operating costs. The electric energy industry in particular makes extensive use of weather information supplied by both government and commercial suppliers. Through direct purchases of weather data and information, and through participating in the increasing market for weather derivatives, this sector provides measurable indicators of the economic importance of weather information. Space weather in the form of magnetic disturbances caused by coronal mass ejections from the sun creates geomagnetically induced currents that disturb the electric power grid, sometimes causing significant economic impacts on electric power distribution. This paper examines the use of space-derived weather information on the U.S. electric power industry. It also explores issues that may impair the most optimum use of the information and reviews the longer-term opportunities for employing weather data acquired from satellites in future commercial and government activity. Dr. Henry R. Hertzfeld is senior research scientist, Space Policy Institute, George Washington University. Dr. Ray A. Williamson is research professor, space policy institute, George Washington University. Avery Sen is a second year student in the Master of Science, Technology and Public Policy Program. INTRODUCTION For 2000, the U.S. electric power industry earned an estimated total revenue of $247 billion. Electric power generation is thus a very large and important industry. Modern society depends on electricity to supply much of the power needed to support manufacturing, and daily heating, cooling, and lighting needs. It is therefore an essential part of the infrastructure of the U.S. economy. Even a very small service disruption can have a large social and economic impact. The 1977 blackout in the Northeast U.S. cost the U.S. economy an estimated $340 million (in then-year dollars); the August 2003 blackout may have cost New York City alone some $1.15 billion (estimates of total cost range from $4 to $6 billion). Neither unusual terrestrial weather patterns nor space weather appear to have caused either of these two blackouts. However, both types of weather incidents are capable of creating major problems with the electric power infrastructure and therefore have the potential for causing large economic losses. Terrestrial weather conditions, typically, are predictable; better forecasts will lead to more efficient management of the electric power system and, as described in this paper, contribute to sizable cost savings. Incidents caused by space weather are not as predictable and can occur within minutes to a few hours of a coronal mass ejection from the sun, but in the last decade, scientists have made measurable progress in understanding the physical basis of space weather and in extending their ability to predict harmful consequences on Earth. Accurate weather information is only one component of the smooth and efficient operation of the large and complex electric utility industry. Accurate weather information is most important when significant deviations in temperature or PAPER DELIVERED AT INTERNATIONAL ASTRONAUTICAL CONFERENCE, BREMEN, GERMANY, SEPTEMBER 2003 -2storm-caused natural disasters are probable. Nevertheless, because the industry is very large, because energy prices are volatile, and because of the high cost of capital facilities for energy production, management, and transmission, improvements in predicting and planning for changes in the weather can result in potential annual aggregate savings of hundreds of millions of dollars for the U.S. economy as a whole. In particular, finer, more accurate satellite weather observations from improved instrumentation, when combined with enhanced weather models, can provide the basis for more accurate, short-term and long-term forecasts. Improved forecasts can potentially lead to significant cost savings in electric energy production. The benefits of better terrestrial weather information obtained from satellite data are not limited to the electric power industry. They extend to nearly all socio-economic activities: household, industry, and government. Space weather has a more limited, but important, effect on society as a whole because it primarily affects technological systems, and especially the electric power grid. This paper focuses on the electric power industry because it is one of the largest users of weather data and is potentially one of the largest beneficiaries. THE VALUE OF TERRESTRIAL WEATHER DATA IN THE ELECTRIC POWER INDUSTRY TYPES OF USES Electric utilities use weather information in several different ways. Different personnel, even within the same company, may manage each different type of forecast and use. They generally use different models and different variables, the value of which may vary greatly. In addition, the relationship between the accuracy of the weather forecast and the eventual economic benefit will be different for different uses. Energy utilities use weather forecasts in the following components of the generation process: 1. Fuel acquisition (both purchasing and transportation of the fuel); 2. Load (demand) forecasting; 3. System planning Planners use all types of weather and climatic forecasts in preparing for the purchase and transportation of fuel. Such information is used to determine where to purchase fuel and how it to transport it to the power plant. The Clean Air Act Amendments of 1990 placed new, more stringent emission requirements on electric utilities than they had previously experienced. Weather anomalies (very hot summers, cold winters, etc.) can affect energy costs and their ability to meet emission standards for their service area. Thus, they have to be very sensitive to possible weather changes that might require last minute energy purchases at high prices, either to meet regulations or to meet energy demand. Utilities use forecasts of 1 to 12 months ahead for this planning activity. Weather factors are also very important in the generation of electricity. Most generating units do not perform as well during periods of high temperatures and humidity as they do during moderate conditions. Thus, short-term forecasts and timely weather observations are critical for the efficient operation of generating plants. Because bringing unused plants into operation during peak periods takes hours and even days, utilities also employ two to ten-day forecasts. Electric load (demand) imbalances are very costly to the electric power companies. Hence, forecasts are very important to the industry. Historical weather data coupled with other economic and social variables account for up to 95% of the accuracy of demand forecasts. Nevertheless, weather forecasts are also critical. The broad geographical distribution of demand and the need in a deregulated world to maximize usage means that utilities both buy and sell power on the market. The industry has a spot market for immediate purchases (in times of high demand the price often reaches very high levels) as well as a day-ahead market. More accurate weather forecasts result in major economic benefits to the industry because utilities can then time their purchases and sales of electricity optimally. Electric utilities also use weather data to optimize the analysis of sales and corporate earnings, to provide better customer service, to optimize new transmission and distribution lines (the choice of construction materials and the placement of lines), and to balance the supply and demand of electricity at regional levels. PAPER DELIVERED AT INTERNATIONAL ASTRONAUTICAL CONFERENCE, BREMEN, GERMANY, SEPTEMBER 2003 -3Additionally, they use weather forecasts to plan for and mobilize resources to meet storm and lightning restoration work. Short-term forecasts enable risk management of surges. This includes terrestrial storms and geomagnetically-induced currents caused by space weather, In short, the electric power industry uses weather information of all sorts for a wide variety of tasks, including risk management, system and capital planning, and trading. It will become increasing important in a deregulated marketplace. EVIDENCE CONCERNING ACCURACY OF FORECASTS AND ECONOMIC VALUE Because the industry uses different types of forecasts for different purposes, the value of forecasts will vary depending on use and type of forecasts (i.e. time-frame). This research is focused in part on exploring the relative economic value that can be expected from efforts to improve weather forecasts. One important question is the economic value that can be derived from moving from a seven day-ahead weather forecast to a 10-day ahead forecast. When the average temperature is between 65 (18o C) and 75 (24o C) the electric power industry benefits little from short-term weather forecasts. The main benefits accrue when temperatures rise above or fall below these temperature. The industry uses a measure of heating degree days (HDD) or cooling degree days (CDD), measured as deviations from 65. The industry experiences greater economic consequences from inaccurate forecasts during extreme periods of heat than during unusual cold periods. Beyond temperature variance and time differences, geography is also an important variable. Weather in the Central American isthmus, for example, is extremely difficult to predict because of rapidly changing terrain elevations and the influence of not one, but two ocean systems. Regions characterized by large temperature ranges and strong seasonal effects not only present more difficult forecasting challenges, but may also be the regions where economic benefits from more accurate weather forecasts are the greatest (at least for some types of uses). Forecasts of temperature have two dimensions: the point estimate of a temperature and the expected variance (as measured by the size of the standard deviation from the forecast). Forecast improvements must encompass both—superior point accuracy and reduced variance. The following expert opinions illustrate the current knowledge about the value of different types of forecasts. A literature survey reveals that: • No national or aggregate estimates exist concerning the value of forecasts; and • Most experts cite examples derived from their experience, usually from one application at one utility company over one time period.