…but the coolest summer nights have warmed by 5 deg. F.
John Christy and I continue to examine U.S. air temperature trends, especially those in summer, and John has recently been looking at “heat wave” statistics.
My interest is in determining how much the urban heat island (UHI) effect has impacted reported warming trends. Last year we published a paper using population density as a proxy for urbanization, and found that about 60% of U.S. urban and suburban warming trends in Tavg (the average of the daily maximum [Tmax] and minimum [Tmin] temperatures) since 1895 in the “raw” (non-adjusted) temperature data could be accounted for by urbanization.
But we also found that relationship largely disappeared by the 1970s, with little warming since then being accounted for by increases in population density.
Landsat Impervious Surface Data
We used population density in that study because the datasets are global and extend back to the 1800s (and even earlier). But the most direct physical relationship to UHI warming would be the coverage of the area around the thermometer by impervious surfaces (IS). Those data are now available at 30 meter resolution from Landsat for each year between 1985 and 2024 (40 years). IS might well reveal UHI effects in cases where population density is no longer increasing but wealth has increased (more air conditioning, Dollar Generals, etc.)
But I’m not going to show IS data today, that’s for another time. I’m only explaining how I got here.
D.C. Urban Warming Trends: The Difference is Like Day and Night
For now I’m examining metro areas (which is what the EPA Heat Wave papers also do), using airport ASOS measurements which is what the National Weather Service and FAA mostly rely upon. These systems are well-maintained since their primary purpose is to support air traffic safety.
I started with the center of America’s universe, Washington D.C. And I also decided that something better than a “heat wave” index was needed.
The heat wave (like pornography) is difficult to define, but you know it when you see it. How many days in a row constitute a heat wave? And how hot do those days have to get? Above the 85th percentile? 90th percentile? Those questions do not have definitive answers.
Also, by choosing a binary variable, there is no gray area available for days that are almost a heat wave (oh, sorry, there were only three days above 100 deg. F, so you didn’t meet the 4-day threshold). Such definitions lead to dodgy statistics, such as computed trends in heat waves,
So, I decided (as a meteorologist) that the hottest days in each month make more sense to keep track of for climate trends. I decided on the average of the 3 hottest daily maximum temperatures in each summer month (June, July, and August) as a potentially useful metric, which is approximately the hottest 10% of the days in the month. This metric always exists, every month, every year, and it always has 3 days. This is good for statistical analysis.
But then I thought, why stop there? What about the 3 coolest Tmax days each month?
Which then led to, “What about the warmest and coolest 3 days minimum temperature (Tmin) measurements?”
So, I started with Washington D.C., Reagan National Airport, which is used by your favorite congresspersons and presidents (as well as the public) to keep track of how hot it’s getting.
The results surprised me. Here are the temperature trends in those different categories. What is amazing is that the coolest summer nights in DC have warmed 10 times faster than the hottest summer days:

In fact, the trend in the hottest days’ temperatures is not even statistically significant, at only +0.12 deg. F per decade, which is just under a total of 0.5 deg. F warming in the last 40 years. No Boomer would notice that in their lifetime.
But look at those nighttime temperatures! The coolest nights have warmed by almost 5 deg. F in the last 40 years. This is clearly dominated by the UHI effect, since climate models tell us that days and nights should be warming at much closer to the same rate.
Now, Washington D.C. might be an outlier for urban areas. I’m just starting down this road, so we shall see. But I’ll bet most people would not have expected these results if they have been watching the local D.C. TV stations’ weather and news coverage.