The reality is that America's railroads are growing (for the first time in 40 years) and the trucking industry benefits from stunningly high technology. There is a strong, competitive rivalry between America's railroads and truckers for market share. This battle is literally changing the face of the economy of the United States. And, increasingly, trucking and railroading depend on weather information tailored to their specific needs.

The Economic Setting

In the last two years, railroad "Mega Mergers" have frequently been front page news in the business press. With the Stagger's Act of 1980 (which deregulated the railroad industry), there has been a revitalization of what had been an industry in decline for 40 years.

Beginning about 1940, truckers began capturing market share from the railroads. The improved passenger aircraft of the 1950's (such as the Lockheed "Constellation"), along with jet aircraft in the 1960's, caused the passenger traffic carried by railroads to rapidly decline, especially for interstate travel. The advent of the interstate highway system accelerated the decline in the amount of freight moved by rail as trucks were perceived to be less expensive, more versatile and more reliable. The bankruptcy of the Penn Central and other Northeast Railroads led many observers to conclude that railroads would go into a permanent state of decline, evolving, at best, to mere haulers of bulk commodities, such as coal and grain. "In the late seventies it seemed inevitable in Washington that the industry would have to be nationalized," according to James McClellan, then of the U.S. Railroad Administration, quoted in Trains magazine.

The Stagger's Rail Act allowed railroads to set their own rates (previously regulated by the now defunct Interstate Commerce Commission), terms of service and service standards. To fully capitalize on this opportunity, it became essential for railroads to lengthen their route systems (so that inefficient car interchanges could be minimized) and streamline their operations in other ways, both internal and external. This occurred in the early to mid-1980's when financially weak railroads were taken over by stronger railroads. For example, the Union Pacific of 1995 was a combination of the original Union Pacific combined with Western Pacific, Missouri Pacific, Kansas Pacific, Kansas-Oklahoma-Texas ("Katy"), and Chicago and Northwestern.

The mergers of the 1980's gave way to the "Mega Mergers" of the 1990's. For example, Union Pacific merged with Southern Pacific (the original Southern Pacific, plus the Central Pacific, St. Louis Southwestern [Cotton Belt], and Denver and Rio Grande Western); the Burlington Northern merged with the Santa Fe; and currently, the CSX is purchasing Conrail, which will then sell some of Conrail's routes to Norfolk Southern.

There is, however, no transcontinental railroad in the United States. It is believed by some that the rail mergers of the next decade will merge the eastern and western railroads. Some believe that the United States could have as few as three long line railroads in ten years.

Until the mid-1980's, railroads were controlled through an antiquated system of "train sheets" and "track warrants." This changed when Union Pacific built its Harriman and CSX built its Dufford control center. With this innovation, vast railroads were controlled from a single room, which made it possible, for the first time, to use weather information efficiently and effectively. All of the Class One railroads except Norfolk Southern control their systems from a single location. It is fortuitous that the advent of the national control centers came on the scene when they did, because railroads were developing new types of business and railcars that made them more vulnerable to weather than at any other time in their history.

Between 1978 and 1982, the Southern Pacific teamed with railcar builder ACF Industries and the steamship company American President Lines to create the "double stack" container car. The double stack made it more timely and economical to cross the United States by rail with containers than for ships to go through the Panama Canal. It takes 20 people to take a train of 25 five-platform articulated cars that hold 250 containers (within which everything from chemicals to VCR's are shipped) from Los Angeles to Chicago. It would take truckers 500 drivers (working in tandem) to haul the same quantity of goods in the same amount of time. However, the double stacks, especially when empty, are highly vulnerable to being blown over in crosswinds.

In 1989, the first joint service between trucks and railroads was launched. Known as "Quantum," the Santa Fe Railroad began shipping trailers for the J. B. Hunt Transport Company on flat cars. This service, now common to most major trucking companies and all major railroads, has been economically beneficial to both industries (see Trucking section of this paper). But again, the trailer mounted on a flat car, is highly vulnerable to blowovers.

Finally, railroads have captured a greater share of the long haul automobile shipping market by developing railcars that can carry 15 automobiles at a time while providing outstanding protection against damage in transit. These railcars, known as "autoracks," are vulnerable to overturning in high winds because of their high profile.

While railroads were regaining their economic health, they were becoming more vulnerable to weather and its effects.

Trucking

The interstate trucking industry was also deregulated in 1980, which allowed greater freedom in pricing and in determining which markets to serve. But, like the railroads, the advent of deregulation brought challenges -- especially since the Interstate Commerce Commission would phase out the "floor" under trucker's rates -- which, when combined with the challenge posed by more efficient railroads and by overnight delivery services such as Federal Express, meant the trucking industry had to become more efficient.

The "reengineering" of the late 1980's and early 1990's created an enormous opportunity for truckers, in the form of "just-in-time inventories." This is explained by Mary Walton in The Deming Management Method: "Having been perfected in Japan, 'just-in-time' production is enjoying a surge of popularity in this country. In this system, supplies arrive as they are needed, so that money and storage are not tied up in inventory."

Given the demands of just-in-time production processes, truckers needed to better anticipate and deal with weather-related delays and problems. No longer is a flood a local event. For example, according to Kevin Williams of the American Trucking Association, the Saturn division of General Motors keeps only two days worth of parts on site. If there are delays or interruptions of more than two days in truck (and, to a lesser extent, rail) deliveries, the plant begins shutting down. A major flood event that closes interstate highways (such as the 1993 Midwest flood), can literally cause factories to shut down and layoffs to occur in businesses 1,000 miles from the nearest high water!

Surprising to many is the fact that the major trucking companies know the location of their trucks while they are on the road to a higher degree of precision than railroads or even airlines. For example, the trucks operated by J. B. Hunt are tracked by GPS and are in continuous communication with Hunt's central facility in Bentonville, AR. If immediate communications are needed with a given truck, it is only a few keystrokes away and, for urgent information, an alarm tone sounds in the cab.

Today's interstate truckers handle deliveries of up to 600 miles (eight hours) entirely on the road. Longer deliveries involve putting the trailer on a rail car, shipping it to the distant railyard that is closest to the truck line's customer, then mating the trailer with a truck cab for final delivery to the customer.

Weather Information Requirements

Phenomena	Rail	Trucking
Tornado	Derailment	Blowover
Hail	Little or none	Dented autos in transit
High winds	Derailment, crossing gate damage	Blowover
High Temperatures	Spoilage, sun kinks, locomotive problems	Spoilage, animal mortality
Low Temperatures	Broken rails, locomotive problems	Engine problems
Lightning	Signal irregularities	Little or none
Flash floods	Derailments, reroutes	Reroutes
Floods	Reroutes, performance penalties	Reroutes, penalties
Snow	Delays due to track clearing	Closed roads
Ice	Locomotive traction, signal lines	Closed roads, traction

With derailments and blowovers, there is the related problem of hazardous material spills.

Meteorological Responses

Instrumentation

According to materials presented at a December 1995 seminar sponsored by the Minnesota Department of Transportation (MnDOT), the first "Roadway Weather Instrumentation System" (RWIS) was installed in the late 1960's on an interstate highway bridge. The RWIS was connected to a sign that flashed "Ice on Bridge" when ice was detected on the roadway. There are more than 300 RWIS systems currently operating in the United States. It is estimated that the benefit to cost ratio of RWIS systems is 2.5 to 1 for the DOT's that use them with a 20 to 1 ratio for society as a whole (i.e., fewer accidents, on time deliveries, etc.). Truckers and motorists derive considerable benefit from these systems.

The Union Pacific has an extensive network of trackside weather sensors which have been installed primarily in areas especially vulnerable to nonconvective high winds. In addition, a number of railroads have installed high water detectors at critical locations as well as rockslide detectors (slides are often indirectly related to weather).

It is interesting to note that the railroads often contend that truckers are unfairly subsi- dized as state and federal governments build and maintain roads and highways while the railroads have to build and maintain their rights of way. Regardless of the merits of this argument, this same pattern has emerged in weather instrumentation for surface transportation. For example, the Kansas Department of Transportation has installed more than 60 weather sensors along state highways but none along railroad rights of way. The KDOT even declines to share its weather information with railroad and aviation interests, in spite of the fact that the Kansas legislature, which established the KDOT, explicitly stated that KDOT's mission was to promote "railways, highways and airways" (in that order) in the legislation creating KDOT. This same pattern exists in at least several other states in the central U.S.

Weather Forecasts

For this purpose, weather forecasts are defined as routine, non-urgent information that can be used to plan day-to-day operations.

Railroads use weather forecasts to plan day-to-day operations, including outdoor maintenance. When extreme temperatures are forecast, "slow orders" will be issued in the morning to reduce the maximum speed of trains where temperature-caused track irregularities are suspected. The lead time for these forecasts is 24 hours. With the advent of the Mega Railroads, it has become possible to route trains around major storms (i.e., the Union Pacific now has three routes from Kansas City to San Francisco). However, reliable forecasts of blizzards, major floods and other synoptic scale storms are needed in the 48- to 72-hour time frame in order for this potential to be fully realized.

Interstate truckers desire lead time in the eight-hour range, because most all of their long hauls are done by rail. Snow and ice are their most serious problems, but high winds (50 m.p.h. or greater) and floods interfere with trucking operations. The shipment of livestock by truck can be affected by high temperatures and high humidity. Most (but not all) perishable goods are shipped by refrigerated truck.

Storm Warnings

For this purpose, storm warnings are defined as non-routine, urgent information that affects safety. An unexpected temperature of 105¡ (if, say, 95° was forecast) is treated as a "storm" by the railroad industry since a sun kink affects safety in the same way as a flash flood or other traditional "storm."

The trucking industry has not yet begun to use storm warnings except in the most general sense. J. B. Hunt, for example, has a monitor tuned to The Weather Channel at all times, which is then used in discussions with drivers and with their safety personnel in the field.

The railroad industry has become extremely active in using storm warnings to improve safety. However, the needs of the railroad industry are quite different than the public's needs. As already mentioned, rapidly changing temperatures can be a "storm." Railroads are not affected by large hail. The railroads only care about a storm if it affects their right of way. If, for example, a tornado parallels a track eight miles away, the railroads will continue operating.

WeatherData has created a "track specific" storm warning service for its rail clients. Warnings are created only for the criteria which affect rail operations and they are "track specific" -- meaning the warnings are defined by track points known to the dispatchers, train crews and supervisors. The warnings are created on a proprietary workstation called SmartWARN which instantly flashes them to the railroad in the exact format requested. For some of WeatherData's rail clients, the warnings are displayed on the SmartRAD system (WeatherData's proprietary weather information display) and routed by computer to the specific dispatcher(s) handling an affected stretch of track through their DigiCon workstations, while bypassing dispatchers handling territory unaffected by the storm.

For example, if a tornado warning is issued for a given section of track, the dispatchers stop trains from entering the warning area. If a train is already in the warning area, it is allowed to exit if there is sufficient lead time prior to the start of the warning (i.e., a warning issued at 6:15 p.m. effective from 6:25 until 6:45). If the warning goes into effect immediately, the train is stopped. If a siding is nearby, the train pulls into the siding. If a blowover occurs on a stopped train, the damage is far less than if the train is moving at 70 mph.

The economics are huge. It costs $2,000 per hour to stop a single train. Because of the density of trains on some of the major rail lines (60-100 trains per day, operating at 70 mph), a tornado warning that covers 15 miles of track for 40 minutes can cause as many as seven trains to be stopped. However, most weather-related derailments cost at least one million dollars with the cost ranging up to six million dollars. Typically, the rail companies are self insured up to $5 million per occurrence, so a weather-related derailment causes a major negative impact on the bottom line, not to mention safety.

The economic payback of this service has been enormous. A proprietary 1995 study for the Santa Fe Railroad comparing a weather provider that merely relayed National Weather Service warnings versus WeatherData's track-specific warnings indicated that for every dollar spent on the NWS service, the payback was $1. For every dollar spent on the tailored service, the payback is more than $35, even though the cost is three times as great. This is due to the accuracy of the warnings (CSI of .89) and their specificity. For example, the number of "false alarms" caused simply by the hail component of NWS' severe thunderstorm warnings is excessive from a rail perspective, so much so, that railroads using severe thunderstorm warnings would typically begin to ignore them after a few months. Note: This is not a criticism of the NWS. Their role is to serve the public, not to create products for a particular company or industry.

This high degree of accuracy and customization is possible because of proprietary technology developed by WeatherData along with the high degree of training provided to its meteorologists. The SmartWARN system is a workstation used to integrate and evaluate numerous meteorological datastreams in real time (similar to AWIPS but tailored to WeatherData's needs and the needs of its clients). While WeatherData's meteorologists can generate fax, telephone or pager-based warnings with SmartWARN, clients that desire an even higher level of monitoring can receive their weather information through the SmartRAD Weather Information System. The people aspect of these services cannot be overemphasized. Meteorologists are carefully recruited and trained, not just in applied meteorology, but in client needs. WeatherData annually takes it meteorologists on "field trips" which allow them to meet clients, see the geography of particularly weather-sensitive locations (i.e., Donner Pass in California), and learn how the forecasts and warnings they create are actually used. WeatherData's storm warning services for railroads have attracted national attention (see the April, 1997, issue of the magazine Rail News).

The "Public - Private" Partnership

In this era of tight federal expenditures, the Public Private Partnership has proven to be an effective way of insuring that American industry receives the tailored services it needs while the National Weather Service concentrates on providing weather forecast and storm warning services to the public. Commercial weather companies are in an ideal position to become intimately familiar with their clientele and to design services specific to their needs.

Nevertheless, there is a role for government to play, which can help leverage the taxpayer's investment and assist commercial meteorologists and their client companies. These include:

• Data-sharing. The Federal Department of Transportation has paid for many of the RWIS sensors in the field. It should require that data acquired with Federal dollars be available to all. Perhaps the NWS could serve as a facilitator in this regard.

• Data enhancement. While the WSR-88D has been a major technological step forward, there are additional ways to enhance its value. For example, the -88D "holds" the NIDS base velocity and base reflectivity until the entire volume scan is completed (a process that takes four to five minutes). These products should be transmitted immediately. We also suggest that the volume scan strategy be changed to allow additional low level scans with fewer complete volume scans when meteorologically appropriate. It would seem that in a tornado situation reducing the "storm total" precipitation product to ten-minute intervals (rather than the current five-minute intervals in severe weather mode) in return for doubling base velocity and base reflectivity scans (to 2-3 minute intervals) would be an excellent trade off.

• Further strengthen the "Public Private Partnership." The success of this partnership should inspire it to be adopted by all elements of NOAA and by the research community.

• Research. The rift between the research community in meteorology and weather forecasters appears to be as wide as ever. For example, the preprint volume from the most recent conference on Weather Analysis and Forecasting (Norfolk, August, 1996) does not contain a single paper on temperature forecasting (there is one paper about an experiment in MOS temperatures using ensemble guidance)! There is little in the way of real world verification with regard to new models and techniques. The current paradigm seems to be to create a new model and "put it out there." Non-model-based forecast techniques are rarely discussed.

• When discussing the role of human beings in the forecast cycle, Brooks, Fritsch and Doswell (same preprint volume, page 524) state, "It is important to note, though, that although the number of large MOS errors is small, errors larger than 30°F occasionally do occur. Thus, one primary role of human forecasters will be ... rescuing the numerical models on days of extremely large errors." The question posed by this author is how? Today's forecasters are taught to rely on the models. If one relies on the models, how will one know 24 to 48 hours in advance, that they are going to be in error by more than ten degrees? What is the technique?

• Stop "adjusting" the models. While this comment often brands one as a Luddite (generally with a remark along the lines of, "If we did that, we'd still use the barotropic."), the amount of continual "tinkering" (word chosen carefully) does not have a favorable benefit to cost. Experience at WeatherData, Incorporated confirmed by recent personal conversations with personnel at the NCEP in Washington and the NWS Central Region in Kansas City indicates that the MRF guidance has had significant biases in the Plains this past winter that are of the opposite sign as during the winter of 1995-96. Edward Deming, perhaps the world's advocates the following (from Dr. Deming. The American Who Taught the Japanese About Quality):

  "One important ingredient of quality is uniformity. It seems almost too simple and insignificant." (p. 39)

  An entire chapter (pp. 67-82) is devoted to the necessity for stable systems in order to improve quality. The current practice where models have a (for example) warm bias one season, are changed, and then have unknown (warm or cold) biases the next season is a formula for deteriorating quality. A minor improvement in RMS error at 500 mb is not necessarily an improvement in the forecast of sensible weather.

Societal Aspects of Weather

Workshop's Main Page

Table of Contents