HomeWeather NewsSummer warming 1895-2023 in U.S. cities exaggerated by 100% from the urban...

Summer warming 1895-2023 in U.S. cities exaggerated by 100% from the urban heat island effect • Watts Up With That?

From Dr. Roy Spencer’s Global Warming Blog

by Roy W. Spencer, Ph. D.

We are now getting close to finalizing our methodology for computing the urban heat island (UHI) effect as a function of population density, and will be submitting our first paper for publication in the next few weeks. I’ve settled on using the CONUS (Lower 48) U.S. region as a demonstration since that is where the most dense network of weather stations is. We are using NOAA’s V4 of the GHCN monthly dataset.

I’ve previously described the methodology, where I use many thousands of closely-spaced station pairs to compute how temperature between stations change with population density at 10×10 km resolution. This is done for 22 classes of 2-station average population density, and the resulting cumulative UHI curves are shown in Fig. 1.

Fig. 1. Cumulative urban heat island effect in different multidecadal periods for the contiguous U.S. (CONUS), June/July/August, for GHCN monthly average ([Tmax+Tmin/2]) temperatures calculated from regression of station-pair differences in temperature vs. population density in 22 classes of 2-station average population density. The number of station pairs used to compute these relationships ranges from 210,000 during 1880-1920 to 480,000 during 2000-2010.

It is interesting that the spatial (inter-station temperature difference) UHI effect is always stronger in the homogenized GHCN data than in the raw version of those data in Fig. 1. The very fact that there is a strong urban warming signal in the homogenized data necessitates that there must be a UHI impact on trends in those data. This is because the urban stations have grown substantially in the last 130 years. A recent paper by Katata et al. demonstrates that the homogenization technique used by NOAA does not actually correct urban station trends to look like rural station trends. It does breakpoint analysis which ends up adjusting some stations to look like their neighbors, whether urban or rural. To the extend that spurious warming from UHI is gradual through time, it “looks like” global warming and will not be removed through NOAA’s homogenization procedure. And since all classes of station (rural to urban) have undergone average population growth in the last 130 years, one cannot even assume that rural temperature trends are unaffected by UHI (see Fig. 2).

Fig. 2. Cumulative growth in population density (PD) 1880-2015 at temperature monitoring stations in four classes of initial station urbanization, calculated by summing the average year-on-year increases in HYDE3.2 dataset population density at individual GHCN stations having at least two years of record in the 20°N to 80°N latitude band, for initial station PD of a 0 to 10, b 10 to 100, c 100 to 1,000, and d greater than 1,000 persons per sq. km initial station population density.

The regression estimates of change in temperature with population density (dT/dPD) used to construct the curves in Fig. 1 were used at each individual station in the U.S. and applied to the history of population density between 1895 and 2023. This produces a UHI estimate for each station over time. If I compute the area-average GHCN yearly summertime temperature anomalies and subtract out the UHI effect, I get a UHI-corrected estimate of how temperatures have changed without the UHI effect (Fig. 3).

Fig. 3. Lower-48 (CONUS) summertime U.S. temperature variations, 1895-2023, computed from GHCN “adj” (homogenized) data, versus those data adjusted for the urban heat island warming estimated from population density data.

The data in Fig. 3 are from my 1 deg latitude/longitude binning of station data, and then area-averaged. This method of area averaging for CONUS produces results extremely close to those produced at the NCDC “Climate at a Glance” website (correlation = 0.996), which uses a high resolution (5 km) grid averaged to the 344 U.S. climate divisions then averaged to the 48 states then area averaged to provide a CONUS estimate.

UHI Warming at Suburban/Urban Stations is Large

The UHI influence averaged across all stations is modest: 24% of the trend, 1895-2023. This is because the U.S. thermometer network used in Version 4 of GHCN is dominated by rural stations.

But for the average “suburban” (100-1,000 persons per sq. km) station, UHI is 52% of the calculated temperature trend, and 67% of the urban station trend (>1,000 persons per sq. km). This means warming has been exaggerated by at least a factor of 2 (100%).

This also means that media reports of record high temperatures in cities must be considered suspect, since essentially all those cities have grown substantially over the last 100+ years, and so has their urban heat island.

Source link


Most Popular

Recent Comments