b'As an oceanographer at the Woods Hole Oceanographictwelve months, were the only game in town, Matt Stein, Salients Institution, I did a lot of research on the water cycle, he explains.CEO and co-founder says.I realised there were oceanic signals that we could use to predictSalient Predictions forecasting provides notably more accu-rainfall on land. I wrote a bunch of papers about them. Im an aca- rate longer-range results. demic at heart, but I realised this technology has direct real-worldWere seeing consistently in the range of 5-15% improve-applications. ment in accuracy and reliability over the next best alternative on Not everyone was convinced. Back in the years before theevery timeframe, says Anthony Atlas, Salients vice-president competition, when he was focused exclusively on oceanography,of business development. Thats kind of a conservative answer his ideas didnt always generate warm enthusiasm from peers orbecause it depends on what were comparing to. If you look at funders.just ECMWF or NOAA, we are up to 25% better on certain lead Frankly, there had been resistance to my publications, justtimes. because its a novel idea. It wasnt traditional and it was difficultFor a bit more anomalous events, like when we have a high to get funded, he says.conviction that theres a big deviation coming, the model gets Still, he was convinced on the concept and the competitioneven more confident. Thats where theres even more separation gave him the push to develop the rest of the forecasting system.between ours and other models because the other models tend To build a forecasting tool, he roped in his two sons to build ato get worse as you get outside the range of normal.machine learning artificial intelligence tool to utilize ocean tem- Not surprisingly, a technology that can more accurately pre-perature and salinity data sets to predict precipitation on land.dict weather weeks and even months into the future is catching We did something completely novel. We used machine learn- the eye of farmers and businesses. ing, which has been slow to be adopted in the weather industry.Opportunities to use this tool in agriculture include improving We do feel like were a front-runner for using machine learning incrop yield models, pin-pointing seed selection decisions, refining this timescale, Schmitt says.crop in-put sales and marketing, energy budgeting, crop insur-Competing against professional meteorologists, Schmitt andance planning, risk assessment in farm financing and insurance his sons shocked everyone. and more. Right from the start, we were way ahead of the professionalI was talking to someone in Brazilthe head of South forecasting companies. By the end of the competition, we hadAmerican sourcing for a major beverage company. He said theyd the best score by a factor of two, Schmitt says.advised one of their barley producers to cut back nitrogen to 100 Schmitt and his sons not only won the competitionspounds per acre because they wanted low protein for beer. [The $200,000 top prize, but they also captured a $50,000 bonusproducer] didnt want to do it. He wanted to put on his regular when they proved their model hadnt just been the most accurate150 pounds of nitrogen. But it ended up being a drought year, so for one year; it was the most accurate in every year for the fullthat excess nitrogen pushed the protein content way too high. decade previous.That mistake cost the grower something like $3M over his entire operation, says Atlas. If only hed had access to a more accurate Full-time Predictions long-range weather forecast, he might have made a very differ-With the successful win to their name, Schmitt took the opportu- ent decision.nity to retire from his oceanography career to dive full-time intoOpportunities exist around the globe. In October 2023, Salient weather prediction, while his sons went back to their real jobs,Predictions was awarded a $3 million grant from the Gates he says. The prize money allowed Schmitt to invest into startingFoundation to integrate their data into crop yield models for East a business, and in 2019, Salient Predictions was officially born.Africa in order to support food security via weather-based seed Salients goal is to provide meaningful, accurate longer-rangeselection and other crop management decisions. Major players weather forecasts to a variety of clients, particularly in the agri- like AB InBev and other food and beverage companies have also culture and energy sectors.expressed significant interest in the technology.We have built up a strong team of meteorologists and dataThe biggest challenge is the leap of faith for companies that science experts, including the winner of the temperature com- are used to doing things a certain way, who assume that the past ponent of the forecasting contest. We have developed sophis- is going to replicate itself in the future, to jump towards working ticated proprietary artificial intelligence models that can runwith a new kind of forecast. It is a new way of thinking, and it can alongside the NOAA and ECMWF models and determine thebe hard to move from the status quo, even when the status quo best blend of forecasts for any location and time lead. No oneisnt very good, says Stein.can match the reliability of our forecasts, says Schmitt. The other big challenge, not surprisingly, is competition.Salient Predictions got rolling at the beginning of the pan- We are ahead of the pack. But the pack is not static. And, by demic. Their first couple customers were both agricultural: BASFthe way, the pack includes our friends at these government agen-and an agriculture company in Brazil.cies, who have big resources. So, our job is to stay focused and Salient has now expanded beyond the two-to-six-week sub- continually improve our models accuracy and reliability so that seasonal timescale of the competition all the way up to one year. farmers and agribusinesses can plan a year ahead for weather Even the government models stop at six months. For six toimpact on crop yield and quality, Stein says.SW32/ SEEDWORLD.COMSEPTEMBER 2024'