Systematic Biases in Manual Observations of Daily Maximum and Minimum Temperature

View More View Less
  • 1 Department of Environmental Sciences, The Pennsylvania State University, Monaca, Pennsylvania
© Get Permissions
Full access

Abstract

The authors demonstrate that manual observations of daily maximum and minimum temperature are strongly biased toward temperatures ending in certain digits. The nature and severity of these biases are quantified using standard statistical methods. Temperatures ending in “0”, “2”, “5”, and “8” are overrepresented in the data, with the bias toward multiples of ten being most statistically significant.

Inconsistencies in the distribution of the data by final digit suggest that biasing toward a temperature T may result not only from misobservations of temperatures T ± 1 but also from misobservations of temperatures T ± 2. Although changes adopted by the U.S. Weather Bureau in 1950 in the rules governing the rounding of temperature observations improved several of the biases, all biases remained statistically significant after the rule revision.

To estimate the potential effect of these biases on the mean and standard deviation of a temperature distribution, biasing simulations were performed on various normal distributions. In addition, it is shown that these biases can affect other relevant climatic statistics, such as the number of days that certain temperature thresholds are reached.

Abstract

The authors demonstrate that manual observations of daily maximum and minimum temperature are strongly biased toward temperatures ending in certain digits. The nature and severity of these biases are quantified using standard statistical methods. Temperatures ending in “0”, “2”, “5”, and “8” are overrepresented in the data, with the bias toward multiples of ten being most statistically significant.

Inconsistencies in the distribution of the data by final digit suggest that biasing toward a temperature T may result not only from misobservations of temperatures T ± 1 but also from misobservations of temperatures T ± 2. Although changes adopted by the U.S. Weather Bureau in 1950 in the rules governing the rounding of temperature observations improved several of the biases, all biases remained statistically significant after the rule revision.

To estimate the potential effect of these biases on the mean and standard deviation of a temperature distribution, biasing simulations were performed on various normal distributions. In addition, it is shown that these biases can affect other relevant climatic statistics, such as the number of days that certain temperature thresholds are reached.

Save