Steve Bochanski, climate risk modeling leader at PwC U.S., describes the state of the art in climate risk modeling, and how the technique is being applied to global supply chains.

Climate risk modeling is the practice of “looking into the future and trying to project how what we think of as extreme weather might change in the future,” Bochanski says. He differentiates if from traditional weather prediction, which is mostly concerned with conditions in the coming days or weeks. Climate risk modeling extends its projections much further into the future — PwC’s version, for example, considers five years to be short-term, 10 to 15 years medium-term, and 20 to 30 years long-term. Some models even extend out to the end of the century.

Modeling begins with projections of temperature increases, an outcome of higher global carbon emissions, then goes on to assess the possibility of changes in precipitation, wind patterns and the jet stream in selected locations around the world. It also addresses the frequency and severity of weather events such as hurricanes, tornadoes, windstorms, drought and heat stress.

For the supply chain, the model hones in on particular sites where clients have manufacturing plants, distribution sites and suppliers. The client’s science team also looks at shipping routes across modes, with an eye toward providing the intelligence needed for users to engage in long-term planning of supply lines.

Outputs are expressed as a risk score, from five to 100. (The model doesn’t go all the way down to zero because “it’s difficult to say that an event will never occur,” Bochanski says.) Mostly important for businesses, it expresses the impact of weather changes in financial terms. “Money is the great equalizer across different types of metrics,” he says. 

Artificial intelligence is proving invaluable in boosting the accuracy of climate models. “The possibilities are endless,” Bochanski says. “We’re all in great ideation mode of how to use AI.”