Urbanization affects the temperature trend of GHCN, Part I: Urbanization characteristics of GHCN stations

From Dr. Roy Spencer’s Global Warming Blog

January 14, 2023 by Roy W. Spencer, Ph. D.

I have published many articles in the past (e.g.: here and here) where I refer to evidence that land surface temperature trends from existing homogeneous datasets have some degree of spurious warming due to the urban heat island (UHI) effect. Although it is believed that homogenization techniques eliminate UHI effects on trends, this is unlikely because UHI effects on trends are largely indistinguishable from warming. up globally. Existing homogenization techniques can eliminate abrupt changes in station data, but cannot correct any source of gradually increasing spurious warming.

Anthony Watts has approach this problem for U.S. temperature monitoring stations by physically visiting the sites and recording thermometer exposure to spurious heat sources (active and passive) and comparing trends from measured devices. placed in good position with tendency from devices in poor position. He found that stations with good location characteristics showed a trend in cooler average temperatures than both poorly located locations and official “adjusted” temperature data from NOAA.

I took a different approach using the global dataset of population density and more recently analysis High-resolution Landsat satellite-based measurements of Global Human Settlement “Building” areas. I have also started analyzing weather station data (mostly from airports) with hourly time resolution, rather than daily maximum and minimum temperature data measurements (Tmax, Tmin) ) normally make up the current global terrestrial temperature dataset. Unfortunately, hourly data stations are fewer in number but have the advantage of better maintenance as they aid in aviation safety and allow to check how the UHI effects change throughout the year. day and night.

In this two-part series, I will review the latest official global GHCN data set (Tmax, Tmin) (Version 4) to see if there is evidence of artificial warming from the effects of climate change. urbanization is increasing over time or not. In the latest version of the GHCN dataset, Tmax and Tmin are no longer provided individually, only their mean (Tavg).

Based on what I have seen so far, I firmly believe that there is artificial warming in the GHCN based temperature data. The only question is, how much? That will be addressed in Part II.

This issue is (obviously) important because if the observed warming trends are overstated, then any inference about the sensitivity of the climate system to greenhouse gas emissions due to Man-made is also overstated. (I won’t go into the possibility here that some part of recent warming is natural, that’s a very different discussion another day.)

What I’m about to show is based on the global stations in the GHCN monthly dataset (downloaded Jan 2023) that has enough data to generate at least 45 years of July data over a 50 year period year, 1973-2022. The start year 1973 was chosen for two reasons: (1) it was when the discrete dataset with hourly time resolution that I was analyzing had a large increase in the number of digitized records (remember that, recording the weather used to be a manual process on paper forms that someone had to digitize) and (2) global urbanization data based on Landsat starting in 1975, close enough to 1973 .

Since the Landsat measurements of urbanization have a very high resolution, one must decide which spatial resolution to use in order to relate potential UHI effects. I have (somewhat arbitrarily) selected average grid sizes of 3×3 km, 9×9 km, 21×21 km, and 45 x 45 km. In the global dataset I get the best results with the 21 x 21 km average of the urbanization data and all results here will be displayed for that resolution.

The distribution results of 4,232 stations (Figure 1) show that only a few countries have good coverage, especially the United States, Russia, Japan and many European countries. Africa is poorly represented, as is most of South America.

I analyzed the respective Landsat-based urban settlement diagnoses for all of these stations, shown in Figure 2. That dataset covers a 40-year period, from 1975 to 2014. Here I plotted the 40-year average of urbanization versus urbanization trends over the past 40 years.

There are a few important and interesting things to note from Figure 2.

1. Very few GHCN locations are actually rural areas: 13.2% urbanize less than 5%, while 68.4% urbanize less than 10%.
2. Almost all station locations have an increase in the number of buildingsand none decrease (this entails destroying all buildings, returning the land to its natural state).
3. The strongest growth is in areas that are not entirely rural and have not been strongly urbanized (see curve fitting data). That is, very rural locations remain in the countryside, and heavily urbanized locations will have little opportunity to grow.

One might think that since most stations are less than 10% urbanized, the UHI effect would be negligible. But the thematic study of Okay (1973) shows that UHI warming is nonlinear, with fastest warming occurring at the lowest population densities, with eventual saturation of warming at high population densities. I have previously showed the evidence supporting this is based on updated global population density data that the greatest rate of false warming (comparing neighboring stations with different populations) occurs at the lowest population densities. . It remains to be seen whether this holds true for “construction” measurements of human settlements (buildings, not population density).

Medium Urbanization or Growth Urbanization?

An interesting question is whether urbanization trends (increasing amounts of infrastructure), or just average urbanization, have the greatest impact on temperature trends? Obviously, growth will have an impact. But what about towns and cities that don’t have an increase in construction, but still have an increase in energy use (generating waste heat)? As people increasingly move from the countryside to the city, population density can grow much faster than the number of buildings, as people live in smaller spaces and apartment and office buildings grow. vertically without increasing their impact on the landscape. There is also an increase in wealth, automobile usage, economic productivity and consumption, air conditioning, etc., all of which can cause more waste heat without increasing population. or urbanization.

In Part II, I will examine how GHCN station temperature trends relate to station urbanization in many countries, in both raw (unadjusted) and homogeneous data. (adjusted), while also looking at how urbanization growth compares to the urbanization average.