Weather-sensitive business operations are primarily reactive to shortterm (three
to 36 hours) local conditions (city, county, state) due to the unavailability
of appropriate predictive data at this temporal and spatial scale. Typically,
the optimization that is applied to these processes to enable proactive efforts
utilizes either historical weather data as a predictor of trends or the results
of continental-scale weather models. However, neither source of information
is appropriately matched to the temporal or spatial scale of many such operations.

While near-real-time assessment of observations of current weather conditions
may cover the appropriate geographic locality by its very nature, it is only
directly suitable for reactive response. A potentially more valuable alternative
is using cloud-scale numerical weather models operating at higher resolution
in space and time with more detailed physics. These weather models may offer
greater precision and accuracy within a limited geographic region for problems
with short-term weather sensitivity. Forecasts based on these models can be
used to aid competitive advantage or to help improve operational efficiency
and safety.

Potential Business Value
As former U.S. Commerce Secretary William Daley
stated in 1998, “Weather is not just an environmental issue, it is a major economic
factor. At least $1 trillion of our economy is weather sensitive.”[1] A more
recent study reported in the Bulletin of the American Meteorological Society
estimates that one-third of private industry activities representing approximately
$3 trillion annually have some degree of weather and climate risk.[2] According
to the National Oceanic and Atmospheric Administration, during the period from
1980 through 2005, the United States sustained more than $390 billion in overall
inflation-adjusted damages/costs due to extreme weather events (i.e., more than
$1 billion in damage per event).[3]

Since these costs are across a wide range of geographic and time scales, consider
the more local and short-term impact of weather events. For example, according
to the Air Transport Association, air traffic delays caused by weather cost
about $4.2 billion in 2000, of which an estimated $1.3 billion could have been
avoided.[4] The U.S. Department of Transportation estimates that about 7,000
people are killed and 800,000 are injured each year in weatherrelated accidents
on U.S. highways. The economic impact of these and other weather-related problems
on the roads is estimated to lead to 544 million vehiclehours of delay and an
economic impact of about $42 billion annually.[5]

Applications to the Industry
Utility companies and energy producers
rely on weather forecasts provided both by government and private meteorologists.
They use this information to determine whether to power up peakers, manage their
assets or buy and sell energy on the world market, which is susceptible to hype
and vulnerable to forecasting errors. It has been estimated that the annual
cost of underpredicting or overpredicting electricity demand due to poor weather
forecasts is several hundred million dollars in the United States alone.[6]

For example, a three-degree Fahrenheit difference between the forecasted and
actual temperature for the Tennessee Valley Authority could result in a 1,350-megawatt
difference in demand. On hot days, that demand must be met by the use of older,
more expensive power plants, which, if used unnecessarily, boost supply costs
by $600,000 per day.[7] Conversely, the local cooling after a thunderstorm can
significantly reduce demand on a hot, humid day, but a utility usually provides
excess electricity based on the conditions before the storm because the forecasts
that they use lack the precision to operate more proactively. Predictions of
precipitation are vital to determining the amount of water available to operate
hydroelectric facilities. Similarly, precise local wind forecasts are critical
to predicting potential power that could be generated by a wind farm and providing
information to determine how equipment should be configured from day to day.

Applications also exist on the distribution side, where forecasts of local
severe weather are important for outage and asset management. These include
preparing for the impact of storm surge, winds and rain on oildrilling rigs
in the Gulf of Mexico from hurricanes; predicting and repairing transmission
facilities that fail from high demand and high temperatures; and scheduling
crews to repair downed power lines due to high winds or accumulation of snow
and ice.

In addition, there is an emerging industry for weather derivatives (as hedges
against weather-related financial risk), which has grown from nothing in 1997
to tens of billions of U.S. dollars today. Initially, this market was for energy-related
commodities but has expanded to other markets like agriculture and retail. While
it focuses primarily on the seasonal scale, it may evolve to include the dynamics
of the short-term market as the local impact of energy commodities grows.

To evaluate the potential benefits of improved weather predictions for the
energy and other industrial sectors, the IBM Thomas J. Watson Research Center
initiated a project to understand the applications of local high-resolution
short-term forecasting. This effort has been dubbed Deep Thunder.

What Is Deep Thunder?
Deep Thunder produces forecasts that provide detailed
four-dimensional information about temperature, winds, precipitation, etc.,
from the surface of the earth to an altitude of about 15 km. This meteorological
modeling effort is unique in the industry and the academic community. This forecasting
capability is designed to be complementary to that of the National Weather Service
(NWS). In fact, Deep Thunder would not be possible without leveraging the investment
by NWS in making data, both observations and models, available. The idea, however,
is to have highly focused modeling by geography with a greater level of precision
and detail while addressing the needs of specific industries, both of which
are outside the mission of NWS.

While one goal of this effort is to improve the technology, another is to understand
the business, safety and other value that such modeling can provide. In reality,
improving the effectiveness of weather-sensitive operations is not really about
the weather. Rather, it is one of optimization of business processes such as
resource allocation, scheduling and routing, which are constrained by specific
weather events. Hence, the value of these predictions will be maximized when
they are integrated into the business processes. Having detailed forecasts of
the right caliber is a critical prerequisite to enable optimization of weather-sensitive
operations.

This effort began with building a capability sufficient for operational use.
In particular, the goal is to provide weather forecasts at a level of precision
and speed to be able to address specific business problems. Hence, the focus
has been on high-performance computing, visualization and automation while designing,
evaluating and optimizing an integrated system that includes receiving and processing
data, modeling and postprocessing analysis and dissemination.

Part
of the rationale for this focus is practicality. Given the time-critical nature
of weather-sensitive business decisions, if the weather prediction cannot be
completed quickly enough, then it has no value. Such predictive simulations
need to be completed at least an order of magnitude faster than real time. But
rapid computation is insufficient if the results cannot be easily and quickly
utilized. Thus, a variety of fixed and highly interactive flexible visualizations
focused on the applications have been implemented to enable timely use and assessment
of the model forecasts.

The initial focus of Deep Thunder was on general forecasting. As the technology
improved and become more practical, other applications were considered. These
include travel, aviation, agriculture, broadcast, communications, energy, insurance,
sports, entertainment, tourism, construction and other industries where weather
is an important factor in making effective business decisions. Essentially,
a further goal of Deep Thunder is to enable proactive decision making affected
by weather, by coupling predictive weather simulations with business processes,
analyses and models. Currently, Deep Thunder forecasts are being produced on
a regular basis for seven major metropolitan areas in the United States at 1-km
to 2-km resolution.

A Mighty Wind
On the distribution side, highly localized weather model forecast data can be
applied to operational decision making in the maintenance, repair and utilization
of the transmission network. Deep Thunder can provide sufficient precision to
enable utilities to plan for power usage, outages and emergency maintenance.
Consider, as an example, the severe wind storm that affected the New York City
metropolitan area on the morning of January 18, 2006. This was caused by a strong
cold front and heavy rain with wind gusts over 60 mph, which led to innumerable
downed trees and power lines. As a result, electricity service was disrupted
to more than 250,000 residences and businesses in the New York, Connecticut
and New Jersey suburbs. In some areas, it took nearly a week for power to be
restored. In addition, there was widespread disruption of transportation systems
(e.g., road and bridge closures, airport delays) and some flooding.

Figures
1 and 2 show just a few aspects of a Deep Thunder forecast for this event. Figure
1 shows a map of predicted sustained wind speeds at 7:00 a.m. on January 18,
2006 for Westchester County, New York, which was severely affected by the storm.
Figure 2 shows both surface and upper-air wind speed and direction as well as
other weather variables that illustrate the onset of the storm at White Plains
Airport in the southeastern part of the county near the Connecticut border over
a 24-hour period.

This operational forecast was available on an internal-to-IBM website before
noon on January 17, 2006; more than 15 hours before the impact of the event
was first observed. Imagine what local utilities and government agencies would
have been able to do if they had access to this detailed and correct prediction
as opposed to other forecasts which did not provide such information. At the
very least, the ability to stage resources in anticipation of this event may
have reduced the lengthy outages that many in the area experienced.

Deep Thunder can also improve generation-side load forecasting by providing
high-resolution weather forecast data for use in electricitydemand forecast
models. Integrating leading-edge data and analytics technology into the operational
decision-making infrastructure of the utilities industry enables a proactive
rather than reactive approach for weather-sensitive business processes. This
idea is illustrated in Figure 3, which shows a screen capture of a prototype
interactive application integrating a Deep Thunder weather forecast with a simple
loadprediction model.

Figure 3 shows a map of Georgia with forecasted heat indices at 8 km resolution.
Major cities and locations of the generators owned and operated by Georgia Power,
the local electric utility, are shown by name. Each power plant location is
also marked with a pin, whose height and color indicate a predicted electricity
demand. A dual encoding is used because the capacities of the power plants range
over five orders of magnitude. Hence, height is a linear mapping while color
bands are scaled logarithmically. The user has the ability to select the type
of power plant (fossil, hydroelectric and/or nuclear), the ability to select
what data to show on the map (e.g., weather, geographic or other customer/demographic)
and the ability to query individual power plants (i.e., by visual selection).
The results of the query include the predicted load at each time step (every
10 minutes) as well as a plot of predicted load over 24 hours with weather data
at that location.

Conclusion
Deep Thunder can be a powerful tool for the energy and utility
industry for use in short-term weather forecasting where precision and speed
are critical factors in making effective decisions. Deep Thunder can help companies
avoid being forced to react to weather events. It can enable their weather-sensitive
operations to be proactive, which will have the potential to aid in competitive
advantage and/or help improve efficiency and safety. For more information about
Deep Thunder, visit http://www.research.ibm.com/weather/DT.html.

Endnotes

  1. Congressional testimony.
  2. Dutton, John A. Opportunities and Priorities in a New Era for Weather and
    Climate Services. Bulletin of the American Meteorological Society. 83, No.
    9, 2002, pp. 1303-1311. http://ams.allenpress.com/pdfserv/ 10.1175%2F1520-0477(2002)083%3C1303:OAPIAN%3E2.3.CO%3B2
  3. Billion Dollar U.S. Weather Disasters, NOAA/NCDC, 2005, http://www.ncdc.noaa.gov/oa/reports/billionz.html
  4. Air Transport Association. State of the U.S. Airline Industry: A Report
    on Recent Trends for U.S. Carriers. Washington, D.C., 2002.
  5. Lombardo, Louis. Overview of U.S. Crashes & Environment. OFCM WIST II Forum,
    4-6 December 2000. http://www.ofcm.gov/wist2/presentations/ day2/5_panel4a/2_lombardo.ppt
  6. Economic Statistics for NOAA, May 2005, http://www.publicaffairs.noaa.gov/pdf/economic-statistics2005.pdf
  7. Patrick Walshe. Role and Impact of Weather at the Tennessee Valley Authority.
    Fourth Annual User’s Forum, 86th Annual Meeting of the American Meteorological
    Society, 29 January–2 February 2006, Atlanta, Georgia.