AI revamps climate forecasts for supply chains

Bloomberg— Himanshu Gupta realizes the havoc he has caused Climate change in agriculture. While growing up in India and working in public policy making, he witnessed how unexpected delays to the monsoon season destroyed crops and degraded the lives of his farmers.

“I knew there was a big problem,” he explains of the risks climate change poses to agricultural products such as cotton and its impact on livelihoods. The question was: Do we have the technology to solve this problem?

This experience prompted him to co-found from my brother, a San Francisco Bay Area startup that aims to help farms and other businesses adapt to a warmer, more problematic climate by turning to artificial intelligence. Using machine learning models, the new company claims that its customers Predictable and prepared for it In the face of climate risks that threaten supply chains and their activities for periods ranging from weeks to seasons. These deadlines, Gupta says, typically pose a dark area for weather forecasters who don’t have AI-based tools at their disposal.

It is difficult to accurately predict severe weather events such as hurricanes and heat waves for very specific geographic areas weeks or seasons in advance. However, there is Adequate time to deal with natural disasters can be crucial in whether they are avoided or not.

If you tell a company’s supply chain manager that a heat wave is coming that could impact the cotton supply chain a week before it arrives, that’s not enough time to move inventory or prepare to move forward, Gupta said. “But if you tell them that there is a greater risk of a heat wave next season in this area where we think the cotton crop is down, that will be very possible for them,” he added.

See also  System incisions under the magnifying glass Raghuram Rajan | newspaper

ClimateAi creates deep learning models, Incorporating ocean parameters such as sea surface temperature and ocean salinityTo better predict climate risks, even for areas with limited meteorological data.

“The secret is that the oceans are the long-term memory of the Earth’s climate that determines the average and long-term climate of a given place, and we can tap into that memory using machine learning methods,” Gupta said.

These methods involve assimilating data from a variety of sources and using it to look for patterns and make predictions. ClimateAi combines its deep learning models with government-run forecasting models, such as those from the US National Oceanic and Atmospheric Administration, and digital models such as those used by the Intergovernmental Panel on Climate Change to achieve what it says is more accurate and effective. Reliability in forecasts.

How can artificial intelligence help with climate change?

The emergence of artificial intelligence has led to an increase in the number of startups using AI tools to help companies and organizations prepare for climate impacts. Jupiter Intelligence, which includes its clients With Edison And baby. (BP), also uses models to analyze climate-related financial risks. Google has a research group that applies artificial intelligence to weather and climate challenges. In April, climate intelligence company, which also makes AI models, became the first private company to launch its own radar weather satellites that it hopes will help improve its models.

But what sets ClimateAi apart, according to co-founder and former CTO Maximilian Evans, is what he calls “impact modeling.” Basically, the company Creates two different types of models that can communicate with each other. One focuses on climate and uses inputs from satellites, radar stations, and data from public and private weather stations. The other is business focused and includes historical customer data for variables such as crop yields. Combining them could provide a clearer view, for example, of the impact of heat risks on crops.

See also  Franklin Templeton warns of a recession in Mexico

Evans said the goal is not just to provide accurate climate forecasts; It’s about providing “decision accuracy” that allows companies to feel comfortable using forecasts to take action.

Most of ClimateAi’s roughly 40 global clients are in the food and agriculture industries, but Gupta hopes to expand further into other sectors, including manufacturing and energy.

Some of the company’s existing clients are already starting to see the benefits of using the technology. Driscoll’s Inc., a leading supplier of fruits and other produce, uses ClimateAi tools to help its operations. The company is one of the world’s largest berry suppliers and generates annual revenues of US$4 billion from its production of strawberries, raspberries, raspberries and blackberries.

Although the berries are grown throughout North America, about half of the company’s farms are located in California, a state prone to extreme heat spells, drought and wildfires. Mulberries are difficult to grow in very hot conditions and are sensitive to climate changes. Strawberries, for example, suffer from discoloration and poor root growth in very hot climates. Drought and water shortage result in small berries and lower crop productivity. That’s why the company is using ClimateAi technology to help Predicting where to locate raspberry farms in the future Defining a long-term growth strategy.

Exploring a new site usually takes many years and involves sending local teams to collect soil and climate data. But with the help of predictive tools like ClimateAi, companies can find new locations “within minutes,” Evans said.

“We’re seeing the impact of climate change. In California we have all this volatility,” said Soren Bjorn, Driscoll’s president for the Americas. “This is like an insurance policy.”

See also  More countries are imposing restrictions due to the new variant of the Omicron coronavirus

Despite the promise, there are challenges that AI models face when predicting future local climate changes.

Data in some parts of the world is still sparse, which could be a problem for companies outside the United States and Europe, said David John Gagne, a machine learning scientist at the National Center for Atmospheric Research. Some sites contain more observations from surface instruments and weather balloons than othersSo training a model using data primarily from the United States and applying it to make decisions for a company operating in South America, for example, could result in a less accurate model due to the difference in climate. .

“There have certainly been improvements in the outlook, but there is still a lot of uncertainty in the process,” Gagne said. “Uncertainty, in some cases, can be greater than would be desired by the people who have to make the decisions.”

Gupta acknowledged that this is a problem for ClimateAi. The company lost a Brazilian client because ClimateAi’s predictive model could not be applied to its farm, a problem Gupta attributed to the low density of weather stations in Brazil.

“From a probability standpoint, there will always be a chance we could be wrong,” Evans says. But the status quo and the alternative are based on historical averages, which increasingly predict what might happen in a future affected by climate change.

“So it’s not about ‘what’s the other side,’ it’s about ‘what’s the better alternative?’” he said.


Aileen Morales

"Beer nerd. Food fanatic. Alcohol scholar. Tv practitioner. Writer. Troublemaker. Falls down a lot."

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top