How to Value Weather Information: the Case of a New Aircraft-based Meteorological Data

Extensive literature is available on the development of an economic approach to value weather data and analysis of the cost-benefit characteristics of alternative meteorological data acquisition systems. However, to determine the economic value of a specific set of meteorological data is still a very complex task. Because the data is used in wide variety of weather forecasting processes, which render various and very complicated economic impacts in the society. The TAMDAR (Tropospheric Airborne Meteorological Data Reporting) system is a new aircraft-based meteorological data collection system. In this system, aircraft transmit weather data as they fly in the troposphere. In order to achieve the economic optimization of the system operation, one has to develop a valuation methodology for the TAMDAR data. Decision makers can use this methodology for selecting the best daily data collection configuration. This paper employs a multi-attribute utility function approach to value meteorological data that TAMDAR will produce. Introduction TAMDAR (Tropospheric Airborne Meteorological Data Reports) is a new airborne weather data acquisition system, which is currently developed by NASA. The TAMDAR concept consists of sensor packages, information processors, and communications equipment carried aloft by participating aircraft. As these aircraft complete their missions, the TAMDAR system transmits meteorological data to ground-based receiving stations that process and distribute this data into a national system for dissemination of weather information. The Federal Aviation Administration (FAA), National Weather Service (NWS), and NASA have concluded that TAMDAR systems have potential to improve the accuracy and completeness of weather data and the resulting weather forecasts. Improved aviation safety, along with other benefits for various groups, is anticipated as a result of these improved forecasts. The primary objective of this research is to develop a data valuation approach for the TAMDAR data that will be used to optimize the data collection process and consequently reduce the operational costs. The next section of this paper covers previous studies on the value of weather information in aviation sector. This section summarizes the findings of the literature search, which has been done to develop a methodology to value TAMDAR data. Third section focuses on the description of the developed approach. In this section, the main elements of the valuation model are described without describing mathematical and technical details of the system. Finally, last section summarizes the findings. Value of Weather Data The main utility of information is to reduce the uncertainty, therefore, it presents an economic value to its users. On the other hand, it is hard to treat information as a simple economic commodity. Because it is often very difficult to quantify its use and also very hard to appropriate. Weather information is seen as a public good. It is free for whole community which supports all of its costs and shares its benefits (Bounfour and Lambin, 1999). A five step approach for valuing a weather data set is shown in Exhibit1. In order to value a specific weather information, one should focus on its primary beneficiaries. The expected primary beneficiaries of the TAMDAR data are aviation sector and its customers. Because the TAMDAR data has significant potential to impact the quality of short term forecasts, which are generally demanded by aviation users. There are various studies on the weather forecast impact on the aviation sector. However, these studies are generally focused on isolated cases or specific applications and locations. Some of these studies have achieved to find dollar values for expected benefits. For example, Rodney, Evans and Rhoda (1997) studied delay reduction due to the Integrated Terminal Weather System (ITWS) Terminal Winds Product (TWP). Operational users of the ITWS products include pilots, controllers, terminal and en route traffic flow managers, airlines, Flight Service Stations, and Terminal Automation Systems. In this study several experts were interviewed to determine the TWP uses and their impact on air traffic flow rates at the airport. According to this study the product has been found to be particularly beneficial during times where the capacity is reduced by a combination of low ceilings and/or visibility coupled with a strong vertical shear of the horizontal winds. Analyses using a deterministic queuing model indicate that the benefit of the TWP at Dallas-Ft Worth is approximately $17M/year. In this study authors used three scenarios and determined the probabilities of occurrence for each one of them. By using the delay reduction rate which they obtained from experts, they constructed the quantitative estimates. Exhibit 1. Five Step Approach to value weather data. In a study conducted by Patton, Halsey and Lunnon (1997), user sensitivity to ceiling and visibility is investigated. Authors also studied its influence on Terminal Forecast Verification. This paper measures the sensitivity by the cost to the user of a wrong forecast. This study constructs a benefit/cost measure based on the fuel loading costs of commercial aircraft. In other words, the cost incurred to airlines in the use of optimistic or pessimistic forecasts of ceiling and visibility at the destination and diversion airports usually occur at the planning stage with respect to fuel loading procedures. The results show that for 77% of the aircraft the cost to airlines of bad forecasts is between ₤69 and ₤107. There are other approaches, which are mainly qualitative, case-based analysis. Crowe, Boorman and Isaminger (1997) conducted an operational impact analysis for Memphis International Airport. This study is focused on the impact of thunderstorm growth and decay on air traffic management in a specific type of airspace. In this study, authors identify five typical cases which impacted air traffic in and around the Memphis Class B airspace. Based on the observations derived from these five cases, qualitative conclusions are presented in this paper. Wilson and Clark (1997) estimated the avoidable cost of ceiling and visibility events at major airports. They considered two airport situations, a realistic situation at San Francisco International Airport (SFO), and synthesized very large airports (LRG 16 and LRG20). They developed a simple queuing model and applied to cases of airport arrival capacity reductions that are typical of many ceiling and visibility impacts. A steady traffic model is used in this study. The costs associated within this study are assigned based on accepted ITWS accounting: $1000/hour for airborne holding and $3,800/hour for delay. According to the results of this study a cost reduction between $100K and $200K per delay event can be achieved by using an efficient weather product at SFO. Defining major weather types (modes), which induce aviation delays is useful in order to define value/utility functions. Exhibit 2 summarizes the major delay inducing weather modes for different airports. As seen in the table, major weather types vary with the geographical location. For example, in Chicago, thunderstorms are the primary cause of the delays while in San Francisco International Airport, thunderstorms are not observed frequently. In order to evaluate the utility of TAMDAR, we can first study the utility of similar data sets. ACARS is the weather data collected from commercial jets. Most of the studies, which focus on ACARS data, rely on case studies and they are qualitative, in other words, they do not provide any quantitative relationships about the cost/benefit values. Exhibit 2. Major delay-inducing weather types in different airports. Airport Major Delay-inducing Weather Type Atlanta Hartsfield International Airport 1. Heavy fog 2. Thunderstorms 3. Reduced visibility Chicago O'Hare International Airport 1. Thunderstorms 2. Heavy fog 3. Reduced visibility San Francisco International. Airport Reduced visibility ( however, climatologically few days with thunderstorms) Newark International Airport 1. Ceiling & visibility 2. Thunderstorms 3. High wind Determine the primary beneficiaries and consumers of the data Determine the weather forecast products, used by primary beneficiaries/consumers and in which the data play significant role Understand how data is used by the meteorologists in their forecasting process Construct multi-attribute utility functions, which reveal the relationship between the value of the data to the forecasters and its different attributes Validate and test the utility models Meteorologists agree that aircraft based weather data present high utility but there is no quantitative description of this value. ACARS data is often available at places far from radiosonde sites, and at asynoptic times, therefore, it should be very useful in forecasting convective weather and other phenomena. Aircraft, which fly near airline "hub" airports (such as Chicago O'Hare) sometimes transmit soundings of wind and temperature every 15-30 minutes during the busiest times of the day. The availability of frequent soundings allows the forecaster to monitor the degree of instability and wind shear throughout the day, and issue improved forecasts of convective initiation, severity, and dissipation. (Mamrosh, 1997) In order to develop an efficient data valuation scheme, one should understand its expected spatial and temporal coverage. Therefore, current aircraft data can also provide a basis for understanding the data coverage dynamics. Jamison and Moninger looked in detail at world-wide and Continental U.S. coverage of aircraft data. They stratified the data by day of week, time of day, and altitude. Their conclusions were summarized as follows (Jamison et al, 2001): • Current ACARS data are highly temporally variable. For example, during weekends, data volume decreases. • Most data are gathered at higher flight levels (25 Kft 45 Kft). • Below about 25,000 ft, data are typically concentrated near major hubs. ACARS has been used as information for initial analyses of numerical forecast models. Even though it is an important data source for the Rapid Update Cycle (a computer-based weather forecast model) and meteorological research, ACARS is relatively unknown to most meteorologists. A few airlines have made this information available locally to a nearby NWS office. United Airlines has provided the NWS Chicago office with this data for the last three years, and it has become an important source of data in many forecast and research situations (Mamrosh, 2000). The abundant ACARS data from the O'Hare vicinity has allowed forecasters to more accurately predict the precipitation type of winter storms, high and low temperatures, lake effect snow and strong winds. (Mamrosh, 2000). According to the recently concluded national evaluation of ACARS data, additional applications by other weather forecast offices include the following (Martin, 2000): • Monitoring of changes in vertical temperature profiles with applications to precipitation-type forecasts • Forecasts of mixing heights for fire weather forecasts • Monitoring changes in stability and low-level vertical wind shear with applications to convective forecasts (Mamrosh, 1998) and storm-type forecasts • Monitoring of winds aloft to maintain a smooth flow of air traffic into major air traffic hubs, avoiding problems associated with "compression" of air traffic into hubs Martin (2000) summarized his study by concluding that ACARS data have the advantage of providing greater spatial and temporal sounding detail in the vicinity of areas with significant commercial air traffic and at times of day when such flights are the most dense. Such data can be a useful supplement to upper-air data from the more widely spaced and less frequently available radiosonde soundings. The major limitations of ACARS soundings relative to radiosondes are the lack of moisture data and spatial and temporal limitations on availability. Few ACARS-equipped aircraft contain moisture sensor instrumentation, hence availability of moisture data is very limited. However, the moisture data that is available is of high quality. Because few commuter aircraft are yet ACARS-equipped and because service of ACARS-equipped aircraft from the participating carriers is limited mostly to major airports (where frequency is greatest on weekdays and during the daylight and early evening hours), there are limitations on the spatial and temporal availability of the data (Martin, 2000). Multi-Attribute Utility Approach In order to value the TAMDAR data, this study uses multi attribute utility approach. First the primary beneficiaries of the TAMDAR data are identified and aviation sector is found to be the primary beneficiary and data user. As a second step, we have determined the weather forecast products, which service the aviation sector. Third step was understanding how TAMDAR data will be used by the forecasters in their weather forecast products. Surveys and interviews were used at this stage. After accumulating enough knowledge, initial multi-attribute utility functions, were constructed. They were designed to reveal the relationship between the value of the data to the forecasters and its different attributes. To test and validate the utility functions output of the data was evaluated by the experts. TAMDAR data utility attributes and their brief descriptions are included below. Spatial coverage attribute. Weather forecasters prefer a homogenous and continuous data coverage. Undersampling and over-sampling (because of data acquisition costs) are not desired and should be avoided. In aircraftbased meteorological data collection, it is impossible to achieve perfectly homogenous and predictable coverage characteristics. This is caused by the fact that the participant aircraft follow their own schedule. Therefore, one should expect to have gaps in geographical coverage. In the developed approach, regions where data points are abundant have lower utilities while regions and locations where data points are scarce have higher utility values. This attribute is designed to ensure the inclusion of spatial coverage concerns in the final computation of utility values. Temporal coverage attribute. Forecasters also need temporally well distributed data. However, aircraft flights are scarce between 12 P.M. and 06 AM. In addition, during weekends the number of flights decreases sharply. Therefore, temporal distribution of data should also be optimized. Temporal coverage attribute is integrated into the utility function by calculating point intensity values in a close temporal neighborhood of each data point. Significant weather attribute. Weather is a dynamic phenomenon and some significant weather events may necessitate higher sampling rates within their proximities in order to develop a better outlook. Significant weather events such as thunderstorms, fog, snowstorms and low visibility conditions have considerable economic impacts on society. Therefore, they should be monitored closely. Accurate prediction of the trajectory of a snow storm may reduce the associated weather related costs. Other meteorological measurement methods (such as satellites) may present incomplete information in such circumstances. Timely upper air data may be crucial in certain cases in order to make an accurate forecast. Therefore, data points, which are close to such events should have higher utility values. Climatologic significance attribute. Some geographical locations may present extra importance for forecasters because of their climatologic characteristics. For example, jet streams play important role in weather events in the Continental U.S. Therefore, forecasters monitor the behavior of this atmospheric phenomenon. Consequently, geographical locations, which are close to this phenomena may have higher priority for sampling. The climatologic attribute feature is designed to integrate this type of attributes in to the valuation process. Priority rating attribute. Some weather forecasting activities require subjective evaluations. Therefore, TAMDAR-DSS should include an attribute, in which forecasters can enter their priority ratings based on their routine assessments. Altitude priorities. Utility of TAMDAR data points is also related to the altitude of the data point. As a widely accepted rule, data points, which are collected from the Troposphere (typically altitudes below 18,000 feet) is more valuable since most of the weather events occur in this region. In addition, most of the forecasters agree that utility of data increases as altitude decreases. Therefore, a utility function for the altitude attribute can be developed based on these observations. Attribute weights. Decision makers assign weights to each attribute to calculate the final utility value of each data point. This is a dynamic process. For example, in a standard calm day they can reduce the weights of significant weather attribute and priority rating attribute. On the other hand, if there is a major weather system in the region, decision makers should increase the weights of significant weather and priority rating attributes. Conclusions This research develops a practical approach for data valuation in aircraft-based meteorological data collection. Although this study is based on airborne sensors, the same approach can be adapted to other weather sensors (surface and/or satellite). No model presently exists that addresses all the important relevant issues that have been identified. The utilities of this study's findings are not limited to weatherrelated activities. The TAMDAR problem, in essence, is a decision-making problem related with information acquisition system. There are various kinds of information gathering systems in today's informationoriented economies such as market data collection activities, environmental data acquisition systems, corporate intelligence gathering, security or safety related information collection, Internet based datamining systems etc. The general framework, which will be showcased in this project, can be adapted to other similar areas. Acknowledgements We would like to thank NASA Aviation Safety Program, Aviation Weather Information Element and the Federal Aviation Administration for sponsoring this research. However, the views expressed in this paper are those of the authors and do not necessarily reflect official policy or position of the U.S. Government.