Trends in Daily Temperatures and the Diurnal Cycle

Daily mean temperatures are typically computed using the average of the daily minimum and maximum temperature. It is well known that there are differences between that approach and computing the daily mean of hourly or 3 hourly temperature observations. Here we show that the different approaches also lead to differences in estimates of the long term trends (1940-2019) of global and regional temperature, using both ERA5 and in-situ surface station data. The differences are largest over land areas in the Tropics, where the trend from daily means is larger than the trend using the average of Tmin and Tmax. Tropical regions are also where trends in Tmin and Tmax differ the most. In the global mean the trend from daily means also outpaces the trend using the average of Tmin and Tmax. We reproduce this behavior in the trends using a synthetic time series with non-sinusoidal diurnal cycle. This study indicates that using the approach of averaging Tmin and Tmax to estimate daily temperatures may be systematically underestimating the rate of global temperature increase in ERA5 due to changes in the diurnal cycle.