Extreme weather, infinite data

In 1900 the Galveston Hurricane led to a storm surge and flooding that killed 8,000.

In 1936 an estimated 1,700 people died across the United States and Canada as a heatwave swept across North America.

Meteorologists and hydrologists across the country work each day to ensure tragedies like these don’t happen today by giving people the information they need to plan and warnings to evacuate. Their mission is to protect lives and save property. Their work is as needed now as it was in centuries past.

According to the National Centers for Environmental Information 2019 saw 14 weather and climate disaster events with losses exceeding $1 billion in the United States. It was the fifth consecutive year with 10 or more $1 billion events.

St. Cloud State graduates are helping to lead the preparation for and recovery from natural disasters like these. They have gone on to work in roles throughout the country.

They are studying the effects of urban heat islands in Phoenix. Working for power companies in California. Defending dissertations at Scripps on atmospheric rivers. And working on projects to control the water levels in Lake Minnetonka to mitigate flooding from Minnehaha Creek in Minneapolis.

Dr. David Novak ’00, who was a SCSU Alumni Association Graduate of the Last Decade in 2008, is director of the National Weather Service’s Weather Prediction Center.

At St. Cloud State University, students continue preparing to be part of the effort in the Department of Atmospheric and Hydrologic Sciences where they are learning how to make models
out of vast amounts of data.

Meteorology has gone from a field with too little data to too much data and that is where the jobs are going — modeling, machine learning and artificial intelligence.

Learning through modeling

Students are learning to model the weather — from a calm summer’s day to a raging storm.

They are learning modeling, so they can learn the software.

Learning computer modeling helps students learn what they can do effectively and where they have sources of error — when they should have confidence in a model and when they shouldn’t.

Professors Dr. Alan Srock and Dr. Tony Hansen have been teaching Numerical Weather Prediction for more than a decade.

Two years ago they updated the curriculum to make it required.

“Most of our students were taking it anyway, and there’s certainly demand in the field for people who know how modeling works,” Srock said. “… The field of meteorology is shifting hugely toward data and processing, analysis and visualization.”

With satellite technology there are millions of observations now that meteorologists need to get into a computer model and have them output something useful.

“That is one of the big challenges in our field right now, and we’re getting better at it, but there’s certainly more to do,” Srock said.

Hansen agreed.

“We adapt the curriculum to meet the needs of the marketplace,” Hansen said. “To meet the needs and to equip the students with skills that will serve them when they need to go out in the world.”

The class is helping students learn modeling by teaching them how they can use the software to learn about the weather and how human actions can influence weather in unexpected ways.

This spring students modeled cases such as the influence changing corn hybrids has on them intensity of rainstorms in the Upper Midwest.

With new corn hybrids, farmers can plant crops in much higher density than in the past, and each plant creates more stalks, which requires more irrigation. With increased irrigation more water is lost through evaporation and ends up as higher moisture content in the air. Does that lead to more powerful thunderstorms?

It’s questions like these students are able to explore through modeling, Hansen said.

They first try running a case to see if it will produce a severe storm, then they go ahead and modify the moisture level to what it would have been in the past to see the difference. A model can’t tell
you everything, but if the moisture levels make a big difference, it’s interesting to note, he said.

Senior Savanna Wolvin took the Numerical Weather Prediction class in 2018 and said it gave her a deep understanding about how a model is set up and how they deal with water and precipitation.

“Most of it was just the math of figuring out how to keep a model from crashing,” she said.

If you run the model too long, it will crash. If you make your grid too course, it will miss key features. If you make your grid too fine, you
add many more grid points and processing takes much longer. It’s a balancing act.

Wolvin, who also served as president of the American Meteorological Society student chapter, chose to model a thunderstorm she’d driven through the summer before. Math is what brought Wolvin to meteorology, and it’s math that keeps her fascinated.

“The atmosphere is calculus. It is three dimensional,” she said. “It’s just a calculus problem, but our computers are only able to do algebra, so everything we do in a model has to be approximated, estimated — because it can only do algebra.”

Because of that limitation, meteorology is always going to have to improve and find new ways to improve the models and incorporate data, she said.

In 2018 Wolvin did a Research Experience for Undergraduates (REU) at Colorado State University. She’d learned modeling in the Numerical Weather Prediction class at St. Cloud State, at her REU she learned how to apply that knowledge to research.

There her project focused on using data analysis to evaluate a climate model on how well it can simulate past nighttime heat waves.

“The best way for us to evaluate how well our models are doing is by modeling the past,” she said.

To test the model she would take it and run it multiple times just tweaking the initial conditions by the amount of a rounding error, which can give many different outcomes.

Different models were giving different outcomes, some were saying the nighttime heat waves were creeping north and western heat waves were creeping east.

But that wasn’t what was seen in the actual data, she said.

Nighttime heat waves are a dangerous form of extreme weather where the heat of the day doesn’t dissipate at night causing physical stress on people and animals.

“At night is when your body can finally rest,” Wolvin said. “You have a heatwave all day long and you’re exhausted, but at nighttime when it cools off it gives you that time to rest and handle the next day.”

But for the elderly, young children and people with heart disease, if they don’t have that cool rest period at night it can be dangerous for their health because their body isn’t as able to cool itself as a healthy adult.

“I loved working on that project,” she said. “It was fun to finally take everything I’ve been learning here and get a chance to apply it early on.”

Wolvin adapted that project for her senior capstone. She is researching models for winter extreme cold to see if they are accurately predicting cold spells.

Extreme cold is still dangerous, but it’s dangerous in a different way than heat waves, she said.

“It’s important to watch out to see if our models have accurately predicted extreme cold in the past,” she said.

Feedback from recent graduates is that the Numerical Weather Prediction class prepared them when for when they went off to graduate school. Many said they were better prepared for running the models than anyone else in their class, Srock said.

“The combination of what we’re doing seems to be pretty good,” he said. “We’re trying to update and stay current as much as we can.”

Hydrologists use models too

Dr. Andrea Thorstensen ’11 works as a forecaster at the River Forecast Center in Minneapolis. The center is part of the National Weather Service and is focused on forecasting river conditions, including flooding for all of the Upper Mississippi Headwaters, the Hudson Bay grand basin within the United States, and the Red River.

She has returned to St. Cloud State as an adjunct professor teaching two hydrology courses a year.

Looking at rivers is much like looking at the weather, it’s all about models and data. Data comes from weather predictions, precipitation reports, river gauges and eye-witness reports.

Today the field is expanding to include more satellite data in forecasting. Satellite information can help identify ice buildup, sensing precipitation and soil moisture levels. River forecast
modeling is also changing. For 30 years river forecasters have used lump modeling that monitors conditions basin by basin, but with the added data available today distributed modeling that uses small grids is being explored, Thorstensen said.

River forecasters use weather models, water models and hydraulic models that predict how water will move once it’s in the river channel.

Thorstensen is responsible for monitoring the headwaters of the Mississippi down to lock and dam three at Red Wing, as well as the Rock River in Illinois and Wisconsin and the tributaries to the Mississippi River and down through northeast Missouri.

Last year she issued a couple of record flood warnings on the Rock River.

A lot is riding on river forecasting.

“If people need to go out and sandbag based on your forecast and it maybe under performs, that town has wasted a lot of resources. On the other hand if you miss an event, that’s where you get the danger to property and lives,” she said.

Researching new ways to save lives and protect property

The Department of Atmospheric and Hydrologic Sciences isn’t just preparing students to work with data, professors and students are also doing research.

Together with research colleagues from the U.S. Forest Service, Srock helped develop the Hot Dry Windy Fire Weather Index (HDW), which is designed to give first responders and emergency management specialists a look at what days are forecast to have more dangerous weather conditions for wildfires.

The goal is to give emergency managers an idea of what days they need to pay closer attention to because of what’s happening with the weather conditions. However, the HDW only helps with part of the fire-forecasting challenge because turbulent weather, topography and fuels, such as trees and grasses, also greatly affect fire activity, he said.

“One of my colleagues said the HDW is kind of like a check engine light,” he said. “I like that analogy. If the index is high, we need to figure out why.”

Right now the index is in its pilot phase. Srock and his fellow researchers are working with incident meteorologists — fire-weather forecasters who get deployed to active wildfires — to see how they use the index and what other needs they can seek to address next.

The index isn’t the answer to all the questions, but it helps get to the bigger picture. The biggest motivation was to develop something that would help the people on the ground to save lives and property.

While researching the HDW Index Srock and his colleagues applied it to data collected from historic fires to see how it performed in settings where they knew what happened and that weather was a key factor. He drew students into this research to help with entering the data and developing graphics.

Students like Jessica McDonald ’17, who was lead author on the second paper published on the HDW. She earned her master’s degree at Texas Tech University and is now pursuing her doctorate there

“She put together all of the graphics,” Srock said. “She’s the wizard behind a lot of the plots we put up on the HDW website.” Srock is enjoying including his undergraduate students in his research — an experience he didn’t have as an undergraduate.

“It’s rewarding to work on research that has real-world benefits with undergraduates in particular who get a chance to be a part of something — that doesn’t happen all the time,” he said. “One of the things I like most about this job is being able to connect students with real-world opportunities and needs.”

It all leads to keeping people safe

Everything meteorologists and hydrologists are doing with all the models and research is about keeping people safe. Wolvin believes meteorology and hydrology are forever going to improve, and practitioners will just keep working toward that goal, Wolvin said.

“They always talk about how we always get it wrong, but we don’t have a snowstorm that kills 50 people like we used to have in the 1800s and 1900s,” she said. “We can watch a hurricane come all the way from close to Africa and make landfall and see it on satellite.

“We can categorize how strong it is before it even hits the ground and give people enough warning to evacuate.”

Given the certainty of future natural disasters, there will be many opportunities for these students and graduates to help.

“We have meteorologists and hydrologists around the state, country and world who apply the skills they learned at St. Cloud State to protect people and their livelihoods,” Srock said. “That’s pretty neat.”