Skip to main content
Menu
Close Search
COVID-19 Updates: Stay Informed

New model to predict weather for S.D. climate divisions

Mike Gillispie at the National Weather Service in Sioux Falls
Mike Gillispie, a senior service hydrologist at the National Weather Service in Sioux Falls, is working on a statistical model that will predict departures from normal temperature and precipitation levels for each of the nine climate divisions in South Dakota six months in advance.

How much rain can we expect and how warm or cool will it be during the growing season? Knowing this information can help farmers figure out which crops and which varieties are more likely to provide the best return on their investment.

National Oceanic and Atmospheric Administration Climate Prediction Center’s three-month outlook is one of the tools SDSU Extension State Climatologist Laura Edwards uses to help producers make these decisions. The predictive tool describes the likelihood of temperatures and precipitation as above normal, near normal or below normal across the nation.

However, Mike Gillispie, a senior service hydrologist at the National Weather Service in Sioux Falls, is working on a statistical model that will predict departures from normal temperature and precipitation levels for each of the nine climate divisions in South Dakota for the next six months. Each climate division covers an average of around 8,500 square miles, or 5.5 million acres.

“I want to help farmers make decisions using a longer time frame … with more details,” said Gillispie, who has personally taken on this project. “It’s something I’ve had a passion about for a long time because producers are so impacted by the weather.”

Gillispie’s statistical model incorporates a variety of atmospheric and oceanic indices, such as El Niño or La Niña, as well as other data. “The model looks at the historical correlations between those indices and what has happened with temperature and precipitation out to a six-month lead time for the different climate divisions,” he explained. For example, using these indices from late March, the model can make predictions for April through September.

Furthermore, the statistically-based model then uses historical data to evaluate the accuracy of its predictions as if the model had been in place in the past. It calculates how many years out of the last 50 or 60 years the predictions would have been correct. If, for instance, the model’s predictions are correct 60% of the time during a 50-year period, that is then incorporated into the development or training of the model, he explained.

“I envision being able to show people based on the statistics and indices what the chances are, in terms of percentage, of seeing a wetter-than-normal summer and planting season, for instance,” Gillispie said. “This will give the end users some confidence—they will know that it is better than flipping a three-sided coin.”

However, he pointed out, “This is not a one-stop shop. It is another tool that can guide farmers and give them a bit more confidence in the decisions they are making.”

Gillispie has been working with the model for several years and now runs the model monthly. “Right now, the outlooks are mostly only being shared internally in the National Weather Service to get feedback and information on how best to present the information in a way that people will find useful,” he said.

He anticipates rolling out the model within a year.  “One of the first people I will work with is Laura Edwards, who has the ag contacts,” Gillispie said. Once Edwards understands what the numbers mean, she can then incorporate the information into the outlooks she distributes through SDSU Extension.

Edwards, who routinely uses NOAA and National Weather Service data, said, “The role of Extension is to take whatever comes out of the research side and translate that into what we call decision support, whether it is flooding, drought or what have you, for farmers and communities.

“What Mike’s pulling together is pretty nice as far as local relevance,” she continued. “From what I’ve seen, this can be a great tool to have in the toolbox to give that local flavor with the climate divisions that have already been identified.”