The objective of this manuscript is to examine the global trend using a current reanalysis data set and then to interpret the regionality of the daily temperature trends by month across the terrestrial Earth. To identify whether the method of calculating the daily trend influences the results, we calculate daily means by averaging the daily maximum and minimum values, and also by considering “daily” as the mean of the 24 hourly values. This approach will strengthen our knowledge of the extent to which nuances in the calculation method influence the regionality and monthly distribution of the temperature trends.

Four hypotheses regarding the daily mean surface air temperature over the terrestrial Earth are tested. The first is that the method of calculating daily mean temperature (i.e., “meanmaxmin” vs. “meanhourly”) produces statistically significantly different temperatures. Testing this hypothesis is important because it has long been recognized that maximum and minimum temperatures change asymmetrically, with minimum temperature often changing more substantially than the maximum (Karl et al., 1993). The next two are that a linear temporal trend exists in the global annual mean temperature and in the global monthly mean temperatures for each of the 12 months, for the Earth as a whole. These hypotheses are important to test again for the post-Satellite Era, now that more updated data and data sets are available. The fourth hypothesis is inspired by the comments of Michaels and Stooksbury (1992) about the importance of the regionality and seasonality of the warming; we hypothesize that the temporal trend is not uniform spatially by month.

Hourly air temperature at 2 m above the terrestrial surface is collected from the ERA5 output for the period 1 January 1981 to 31 December 2020, at a resolution of 0.1° x 0.1° (or approximately 11.1 km at the equator). This data set is compiled from NetCDF files, resulting in an array of 365.25 x 24 x 40 temperature values for each of the 2,212,863 grid points located over land. The mean daily temperature is then calculated, by grid point, in two separate ways: first, as the mean of the maximum and minimum values on a calendar day (“meanmaxmin”), and second, as the mean of the 24 hourly observations (“meanhourly”). The time series of monthly mean global temperatures, calculated from the daily means, is then computed using both the meanmaxmin and meanhourly approaches. A statistically significant difference in the temperature distributions indicates that both approaches should be used in testing the hypotheses.

The first hypothesis – that the two methods of calculating daily mean temperature yield different temperature records – was partially confirmed. The meanmaxmin approach yields statistically significantly higher daily temperatures than the meanhourly, on a global basis, but with significant regionality. Northern Africa and central Eurasia actually show higher temperatures from the mean hourly calculation, with most of the rest of the world showing the opposite. However, the spatial distribution of statistical significance in spatiotemporal temperature trends resulting from the two approaches is similar. This result is important because the instrumental and modeling capabilities that now permit the computation of the mean daily temperature based on data collected throughout the day may not create as much of a step change in apparent spatial distribution of the warming signal as might be assumed, despite the fact that the terrestrial temperatures tend to be higher when daily mean temperatures are calculated as the mean of the daily maximum and minimum values, except in northern Africa and central Eurasia.

While the meanmaxmin or meanhourly approach yield the same spatial patterns, the meanhourly method might be considered to be more accurate because of its additional number of hours in the calculation. However, researchers selecting this method should acknowledge that its use yields higher temperatures in northern Africa and central Eurasia and lower temperatures elsewhere across the terrestrial Earth. Regardless, it is cautioned that all of the results herein may occur at least in part because of the processing algorithms in the ERA5 output.

The second and third hypotheses – that spatially-weighted global terrestrial mean daily temperature values display significantly increasing trends in the annual cycle and in each month of the year – were also confirmed. Global terrestrial temperatures have increased from 1981 to 2020, by approximately 0.026°C yr-1, with a range of approximately 0.022°C yr-1 in August to about 0.033°C yr-1 in October. These terrestrial trends are likely dominating the global temperature trend that are frequently reported.

The fourth hypothesis – that the warming is uneven geographically – was also confirmed. Several “hot spots” of particularly high concentrations of grid points reported significantly increasing temperature trends. The strongest concentrations of warming are in the transition seasons in the Arctic, in July in the Northern Hemisphere mid-latitudes, in Eurasia in spring, Europe and the lower latitudes in summer, and the tropics in autumn. Cooling has occurred in some places at some times of the year, but in general, cooling rates are more likely to be statistically insignificant than warming rates.

Future research should be conducted to attribute causes to the observed concentrations of changing temperatures based on atmospheric and oceanic circulation-based forcing. Continuing research using the most current and updated data will shed new light on an environmental situation that is of keen and urgent interest not only to many natural scientists and social scientists, but also to stakeholders in the government and private sectors, and to the general public. It is hoped that future work will also address another limitation of this study by examining non-linear and cyclical temperature trends.

Furthermore, future research is needed to identify spatiotemporal trends in the third category of warming as described by Michaels and Stooksbury (1992) – the distribution of the warming in the day-night cycle. Such results would assist in identifying the main implications of historical warming. Specifically, temporal increases to daily extreme minimum temperatures, typically observed in early morning hours, would have major implications on sectors such as agriculture (e.g., growing season length), entomology (e.g., insect proliferations), epidemiology (e.g., vector-borne illness), energy consumption (e.g., heating of buildings), and transportation (e.g., road and bridge closures due to ice). Likewise, any observed temporal increases to daily afternoon/maximum temperatures would likely impact human health (e.g., heat stroke), energy consumption (e.g., air conditioning), and agriculture (e.g., increased water demand and drought). Improved understanding of these primary weather/climate impacts will assist in planning for future impacts of extreme weather and climate.

To see the results in detail and read more about our peer-reviewed publication on this research click on the link below:

Historical global and regional spatiotemporal patterns in daily temperature

Rahim, M.A., Rohli, R.V., Mostafiz, R.B., Bushra, N., and Friedland, C.J. (2024). Historical global and regional spatiotemporal patterns in daily temperature. Frontiers in Environmental Science, 11, Art. No. 1294456. doi: 10.3389/fenvs.2023.1294456