How heat island research can improve weather forecasts

Post by Christopher Kucharik

All urban regions on Earth have an urban heat island, where the density of asphalt, cement and other built surfaces causes higher air temperatures than in non-urban areas. This is true even in very small cities in far northern locations, like Barrow, Alaska. Our research team has been particularly interested in how the urban heat island impacts the Madison region.

We have studied Madison’s urban heat island effect in great detail over the past four years, looking at how areas across the urban-to-rural scale experience it. We’ve explored the influence of factors like the time of year or day, snow cover, extreme temperatures, and meteorological conditions, such as wind speed and cloud cover, on the magnitude of temperature differences.

Our extensive and detailed data could now be used with confidence to help predict how daily high and low temperatures will differ and change throughout the day and night across the region at a fine resolution—that of a few hundred meters, or a few city blocks. These data can also tell us something about weather forecasts.


People enjoy springtime temps in the UW-Madison Arboretum on Mother’s Day 2016. A freeze warning threatened shortly thereafter. (Photo by Jeff Miller/UW-Madison)

While some forecasters might hint at small-scale temperature differences across the Madison area, or so-called “microcasts,” they do not offer much of an explanation as to why it might be warmer in, let’s say, Middleton than Oregon on any particular day.

Moreover, forecasts often predict the weather for a general region—for example, Madison’s forecasts are typically based on averages, or the conditions expected to be experienced by the majority of the population.

Furthermore, official hourly and daily weather reports for Madison are recorded at the Dane County Regional Airport. The problem is conditions at the airport, which is in a low-lying and mostly non-residential area, tend to be cooler than in denser urban areas, making it not representative of the whole region.

Thus, not only are forecasts missing the mark, but official records of Madison’s weather are also not representing the entire story. They fail to capture the fundamental driver of variations in daily temperatures across the Madison area: the urban heat island.

If forecasters were to incorporate the impacts of the urban heat island, the general public would get a more accurate picture of how widely daily temperatures can vary, whether you live in downtown Madison, in the less dense Hilldale area, along rural County Highway M or in the suburban Bergamot golf community near Oregon.

To the average person going about their daily lives, these details might not make much of a difference. But there are important examples when forecasts that don’t account for the urban heat island effect could be misleading and have a significant impact on segments of the population.

At the time of writing this blog entry, around May 13th, the Madison area experienced an abrupt weather shift, what meteorologists call a frontal passage, which brought rain and cold nighttime temperatures for two nights. The forecast was calling for nighttime lows of 30-32ºF, and the National Weather Service issued freeze warnings.

This was not good for folks who had already put in their tomato and pepper plants, or had gotten a head start and planted their annual flowerbeds. For the die-hard gardener, the threat of a freeze might put them in a panic to protect sensitive plants from the potentially damaging temperatures.

I wish I could have told the worried urban gardener a different story. If you have a backyard garden in Madison’s densest area, the Isthmus, you probably needn’t have worried. Temperatures there are likely to be several degrees warmer overnight because of the urban heat island.

A look at some data shows what actually happened across the Madison area during this cold snap.

Unfortunately, in some places, the weather forecasts were right on. The Dane County airport, for example, hit 32ºF at 11:36 p.m. on May 14th, tying a record low for the day, and then bottomed out at 30ºF at 2:48 a.m. on May 15th, also tying a record low for that day.

This was not good for the local wine industry. Wollersheim Winery, which is northwest of Madison in rural Prairie du Sac, saw temperatures that reached 27ºF and remained below freezing for several hours, leading to significant plant damage. Other local wineries also reported damage.

However, this is only part of the story. Despite the freeze warning issued, a significant portion of Madison’s population did not experience a hard freeze due to the urban heat island.


The figure above shows temperature data from four weather stations that are a part of the Weather Underground network, a publicly accessible collection of data uploaded by volunteers who have weather stations.

I picked four stations to represent locations that are both affected and unaffected by Madison’s urban heat island, and which give results parallel to those from our own 150-count temperature sensor network. Two stations are located in the downtown Isthmus area (KWIMADIS109 and KWIMADIS084), another is on the west side of the UW-Madison campus (KWIMADIS035), and one is my personal home weather station, which I use for teaching and research and is located in a rural subdivision near Token Creek, west of Sun Prairie (KWISUNPR04).

The most interesting part of the temperature data comparison occurs from after sunset on May 14th through the early morning hours of May 15th.

The two downtown weather stations mirror each other, showing a slow decline in air temperature after 6 p.m. and reaching a nighttime low of 38ºF around 6:00am. Thus, no freezing temperatures occurred, and residents with sensitive plants were probably happy to wake up and see an absence of frost.

A bit further west, the UW-Madison site was several degrees colder, reaching a minimum temperature of 34ºF around 6:00am. While not a hard freeze, there was probably a bit of frost scattered around; although, the time actually spent at 34ºF was limited.

In my rural backyard 10 miles northeast of downtown, temperatures fell from 43ºF at sunset to about 32ºF by 10pm. Overnight, temperatures stayed below 32ºF from about 11 p.m. to 6:30 a.m., bottoming out at 27ºF. These temperatures are a good representation of other low temperatures experienced by rural regions outside of Madison.

My home weather station agreed the best with the airport’s temperatures and was significantly colder than the downtown and campus locations. In fact, the difference between low temperatures across these stations was a whopping 11ºF!

So, the lesson for this particular event was that locations within Madison mostly escaped the effects of the hard freeze, which the weather forecasts did not suggest.

Moreover, the rapid decrease in temperature at the rural station compared to the urban stations illustrates how much easier it is for places that have more grass, trees, and crops to lose heat at night, since they don’t store heat in any amount comparable to the concrete and brick buildings of a city.

In this particular case, rapid cooling in rural areas was not a good thing, especially for those with sensitive plants. But in the case of extreme heat in the summertime, the ability to lose heat more quickly at night provides a definite benefit: relief when we are sleeping.

Our second example of when incorporating the urban heat island into forecasts matters is during heat waves, or consecutive-day periods when the daily high and low temperatures exceed particular thresholds. Daytime temperatures may reach 95-100ºF, and nighttime lows are often greater than 80ºF. Although this is pretty rare in the Madison area, it did last happen in July of 2012.

During heat waves, people living in the densest urban areas are at a greater risk because of the urban heat island.

While the National Weather Service does take into account the impact of urban areas when issuing extreme heat advisories, being able to forecast the exact temperature differences between rural and urban regions around Madison due to the urban heat island is more problematic (but getting easier now that we have data). Weather prediction tools that can detect the effects of urban areas on extreme temperatures are not often used in medium-sized cities, such as Madison or Des Moines, even though these types of cities contain a significant fraction of the US population.

To illustrate, during the abnormally hot summer of 2012, 39 days with high temperatures at greater than 90ºF were recorded at the Dane County airport. However, our urban heat island research showed that downtown Madison experienced 49 days greater than 90ºF—10 days more than the airport!

The main point to all of this rhetoric is that we now have much more detailed information on how local temperature patterns across the Madison region are influenced by the existence of all of us living here. This information should be integrated into weather forecasts, particularly during times of year when temperatures have the biggest impact on our daily lives.

Groups that would benefit from more accuracy in temperature forecasts include people who garden, work in landscaping, make decisions on applying road salt in winter, make energy demand forecasts, or are sensitive to extreme heat due to health issues. In the coming months we will be working to make our information widely available, so that it can benefit the nearly 400,000 people who live in the Madison region.

Weather forecasting has come a long way in the past few decades, with accuracies greatly improved by more computational power, more observations, and more knowledge of how the atmosphere works. But until forecasts account for the temperature impacts of urban environments, where 81 percent of Americans live, they will continue to miss their mark at crucial times.

In the Madison region, we can start making local forecasts more accurate by using our knowledge of the urban heat island.

Kucharik is a professor of agronomy and environmental studies at UW-Madison, as well as the lead principal investigator of the Water Sustainability and Climate project.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s