Summers in the South are hot, but just how hot?

You might look at the temperature for the day on your phone or listen to the news to find out what to wear, but that information is usually coming from a weather station that might not accurately represent what you, a human walking around at ground level, experiences.

It is so hot today.

This is especially true in cities.

The location of current weather stations, which is how we get air temperature, they're often actually not located within a city.

They're more located in airports.

What that means is we don't actually know what the temperature in a very urbanized center is.

And that's especially important when it comes to heat.

A lot of people call heat stress the silent killer because not a lot of people really think about it as a weather-related disaster.

But when there is a heat wave, a lot of people can die from it.

Researchers in North Carolina are trying to paint a more detailed picture of what humans experience when it comes to heat in urban environments by measuring air temperature at around 2 meters or 6.5 feet off the ground.

They're doing that with these tiny sensors, starting with Durham, North Carolina.

Durham is a good place to look at urban heat stress because there's such a wide variety of urban forms.

So we have parks, we have really built up city centers with a lot of concrete, and there's also a wide variety of socioeconomic categories.

So we can really do a lot more meaningful work.

This study used 41 sensors spread across the city to cover a variety of different location types based on land cover and solar impact.

We had two different land cover types.

So we had vegetated land cover, and we had impervious land cover, which is essentially surfaces such as concrete and asphalt.

And then the other two categories within that were shaded and unshaded.

This is one of the vegetated shaded sites near Indian Trail Park in North Durham.

So Christian has just connected to the sensor via the Bluetooth link on his phone.

It's showing temperature, relative humidity, heat index, and the dew point.

So all measures of air temperature.

This is a radiation shield and it's initially what you see with the sensor.

But if we actually take a look inside, if you look in this little window here, that is the probe.

So the air in this little window here is what we're measuring.

Now, if we didn't have this radiation shield on the sensor in places where it's not shaded, like this location, what you'll actually be measuring is the heat from the sun heating up that temperature probe in the middle.

So it's really important for us to have this to protect it and also to make sure that it's measuring air temperature.

And looking at what the probe is currently sensing at the site, there's evidence of the differences in what a human would experience versus data coming from the regional weather station.

It's 73.8 degrees Fahrenheit, according to the sensor.

But if you look at your weather app, it's currently 78 degrees.

So this is a little bit more of an in-depth measure of the cooling effect of shaded or vegetative shading.

The same morning at an unshaded site surrounded by impervious surfaces, the opposite was true.

We're currently at the Haytai Cultural Center.

We're actually about like one minute outside of downtown itself.

In contrast to the vegetative shaded, we consider this one impervious unshaded.

We've driven only for about eight minutes drive to this spot and already it just feels a lot hotter.

So the temperature at this site is 81.7 degrees Fahrenheit, which is interesting because the weather currently says that it's 80 degrees now.

I mean, these metrics basically show that the urban heat island effect is very much real.

Lack of shading and ultimately lack of vegetation does induce the urban heat island effect.

The fact that you have some areas which are warmer than others and those aren't being represented by the current data, it just, yeah, it means that it's something that we really need to understand a lot better.

The sensors recorded air temperature every 10 minutes for two months.

Back at the lab, the team is digging into all that data to paint a more detailed picture of heat impacts across the city.

Some of the things we're seeing were things that we expected.

The vegetated surfaces are cooler than the impervious surfaces.

If we kind of zoom into the map a bit, you can see that one of the cooler areas here, this is located in a really forested, vegetated area.

If we find the warmer points here, for example, they're located in an area with a lot of roads and a lot of buildings and not really a lot of green spaces or parks around.

Looking at average minimum and maximum temperatures reinforces the effects of vegetation on heat experienced.

This is the maximum temperature.

The Haytai Centre is 92.4 versus the Indian Trail Park, which is 87.4.

So it's quite a bit warmer at the Haytai Centre.

But there are a few surprises as well, especially looking at night-time temperatures.

If we look at the night-time area, which is this left-hand side here, the black line is actually higher up than the yellow line, which means that the shaded surfaces are warmer during the night-time than the unshaded.

Which we believe is down to the fact that when the ground heats up during the day, during the night-time, it radiates that heat out and it's actually trapped under the shaded canopies.

That's an interesting result, which could mean something for night-time temperatures.

Especially since hotter evening temperatures have been shown to increase negative health impacts related to heat exposure.

The sensors have added to what we know about Durham's heat landscape, but they still only represent their specific locations.

Using machine learning allows the team to paint a more complete picture.

The 4DE sensor is just a starting point.

With this 4DE sample site as the training data, we use machine learning models and combine it with the remote sensing data to build a machine learning model which can predict for the area we don't have a sensor covered.

The remote sensing data comes from satellites and shows land cover type at a scale of around 100 meters square.

Here you can see a satellite photograph paired with the remote sensing data that's used to tell the model what kind of land cover is present in an area.

On the left side, this is our machine learning model for our max temperature of July and August last year.

And on the right side is the land cover types.

These two map patterns are kind of similar and that is true because the land cover data is actually part of the input variables for our machine learning model.

Since land cover is known for the whole city, the knowledge gained at sensor sites can tell us about other parts of Durham.

This is the boundary of Durham and all those yellow dots are where we install the sensors.

When we bring in the machine learning and remote sensing data, we're able to extrapolate from these 40 sensors to all over the place in Durham.

Remote sensing is helping us to fill in the gaps and figure out what happened in the area we couldn't measure.

Why do we need to know how parts of a city are experiencing heat in more detail?

It kind of comes down to how policymakers can actually, you know, use this data and shape new urban policies or environmental policies for the state to, you know, counteract areas that might be experiencing more heat stress compared to others.

By working with local organizations to determine sensor sites and incorporating socioeconomic information to interpret the data, the team is hoping what they collect and extrapolate can have meaningful impact for people living in the city of Durham.

Cities are less than two percent of the earth's surface, but they're where the majority of people are concentrated.

So by 2050, about 70 percent of the world's population is going to be in cities.

So this is why it's really important for us to actually know how the people in cities are going to be affected by heat stress now and also in the future.