Lloyd Treinish leads the environmental science team in the Industry Solutions Department at the IBM Thomas J. Watson Research Center. A co-developer of IBM's Deep Thunder precision agriculture system, he contributed this article to BusinessNewsDaily's Expert Voices: Op-Ed & Insights.
As the saying goes, everyone complains about the weather, but nobody ever does anything about it.
People may never be able to actually control the weather, but advances in a set of key technologies are helping to close in on what may prove to be the next best thing. That is: The abilities to both predict the weather with pinpoint accuracy and, for the first time, to foresee precisely how the weather will impact supply chains, facilities and other key operations.
For business managers and public officials alike, clearer foresight translates into better preparation, less down time and avoided business disruptions. No single technology is driving this revolution. Rather, incremental gains in a handful of areas are reinforcing one another.
The timing is certainly right. The toll from weather disruptions has been rising steadily. The eleven largest weather disasters in the United States in 2012 led to more than $110 billion in economic damage and 377 deaths, chopping 0.68 percentage points off of U.S. gross domestic product (GDP), according to reinsurance giant Swiss Re. For every dollar of economic damage, only 54 cents was insured.
Predicting weather and its impact
To be sure, forecasting is a familiar tool for many businesses. Farmers, airlines and other weather-exposed businesses have always kept a close eye on the skies to guide both near- and long-term decisions.
The rise of more advanced analytical weather tools is new, however. Those tools are helping nontraditional businesses discover and act on extremely near-term weather events that can shape sales.
Consider a pizza-delivery chain. Weather events can have the same effect as a big sporting event, driving up sales, since families tend to stay home and order in when the outside gets rainy or snowy. Yet no one orders pizza two days in advance. That's why some chains are investing in tools to better manage their just-in-time supply chain. With a precise prediction that, for example, rains are due during the dinner hour, a chain manager can stock up on raw ingredients, and book more drivers to be ready for higher-than-normal demand. Given the risks posed to life and property by power outages, it is no surprise that utilities are among the first to explore how hyperaccurate forecasts can improve their day-to-day operations.
Utilities, such as Detroit-based DTE Energy, are collaborating with IBM, providing in-house data about their grid assets to help Big Blue's Deep Thunder system better model the risks posed by severe weather on power lines, transformers and other grid assets. Deep Thunder is our Big Data analytics technology, for local, customized, high-resolution and rapid weather predictions, and it is helping utilities map out and pre-emptively respond to the threat of big storms before they hit. Consider Tropical Storm Irene in 2011. Two days before the system made landfall, Deep Thunder calculated that it would weaken to a tropical storm with greater harm from flooding rains, rather than high winds.
For utilities in the area, this clearer foresight helped guide the pre-placement of the right sorts of repair crews, with the right replacement gear, to the highest-risk spots before the storm hit. Smarter preparation can shorten the duration of outages to hours from days.
Better forecasts: Why now?
The steady march of technology helps explain why these gains are possible. No single advance is revolutionary. But blended together, they're delivering a virtuous cycle of progress.
Bigger, better data. As recently as a few decades ago, many weather measurements were sparse and often manual, tracking a small number of key variables such as temperature, pressure and humidity — and doing so just a few times a day. In recent years, the falling cost of sensors and communications has led to an explosion of more instrumentation with greater accuracy taking more frequent measurements across many different types of metrics. By one estimate, more than 20 terabytes of weather data are available to our national forecasting centers, daily. Deep Thunder goes even further, drawing on data from sources not typically used by forecast services, such as NASA spacecraft, Earth Network's private WeatherBug stations, and thousands of public ground sensors.
Advanced theoretical models. Without better models, bigger, better data are useless. But mathematically complex weather models have been evolving for more than a century. The deluge of direct-sensing data has accelerated the evolution of these complex programs, as new types of measurements, recorded more frequently, help researchers to continuously refine their models.
Higher performance computing. Hyperlocal weather predictions can reveal very big differences between relatively proximate areas. In New York, for instance, the effects of a snowstorm can be mitigated by ocean warmth more readily at John F. Kennedy International Airport than at LaGuardia Airport or Newark Liberty International Airport. By chewing through more data, improved physics models, and more efficiently leveraging supercomputer systems, Deep Thunder is able to zoom its predictions into tinier areas than conventional forecasts and render detailed predictions from three hours to three days out. On the ground, that's the difference between a forecast customized for Central Park versus one for all of New York City.
Incorporating risk. The final piece of this predicative puzzle comes together with data drawn from a previously unconnected realm of property damage. Insurers, public authorities, utilities and other businesses know a lot about how property endures severe events, yet weather forecasters typically know very little. By feeding results from hyperlocal weather forecasts into a linked model of how key assets will respond to a weather event, managers can make critical decisions sooner.
A business problem
To be sure, no amount of computing power will ever "solve" the weather — it is among the most chaotic problems that scientists have ever modeled. Yet those efforts are bearing fruits. Thanks to better predictions, weather is no longer just a problem "that nobody does anything about." Rather, with the right tools, it's becoming a business problem that can yield competitive opportunities.
The views expressed are those of the author and do not necessarily reflect the views of the publisher. This version of the article was originally published on BusinessNewsDaily.