Sampling Biases in Datasets of Historical Mean Air Temperature over Land

Global mean surface air temperature (T(a)) has been reported to have risen by 0.74°C over the last 100 years. However, the definition of mean T(a) is still a subject of debate. The most defensible definition might be the integral of the continuous temperature measurements over a day (T(d0)). However, for technological and historical reasons, mean T(a) over land have been taken to be the average of the daily maximum and minimum temperature measurements (T(d1)). All existing principal global temperature analyses over land rely heavily on T(d1). Here, I make a first quantitative assessment of the bias in the use of T(d1) to estimate trends of mean T(a) using hourly T(a) observations at 5600 globally distributed weather stations from the 1970s to 2013. I find that the use of T(d1) has a negligible impact on the global mean warming rate. However, the trend of T(d1) has a substantial bias at regional and local scales, with a root mean square error of over 25% at 5° × 5° grids. Therefore, caution should be taken when using mean T(a) datasets based on T(d1) to examine high resolution details of warming trends.