Errors in Early Temperature Series Arising from Changes in Style of Measuring Time, Sampling Schedule and Number of Observations

Abstract

Study of the Padova series (1725-today) is a useful example, of general interest, of a critical revision of long time series. These are composed of a number of inhomogeneous parts, each of them with mean daily values, and extremes, computed in different ways, based on observations taken at different times, or with the time expressed in different styles. Imprecise clocks, little care for the schedule established for meteorological readings, changing style of evaluating time, inappropriate choice of observing schedules, too small a number of readings to compute the daily average, generated errors that caused significant departures in time series, that could be interpreted as a climate signal. In the past, average values were obtained with only a few daily measurements. The first problem is to correct the data and extrapolate the hourly temperatures needed to evaluate the daily minimum, maximum and average values in a homogeneous way. The change of style in temporal reference introduced spurious seasonal changes. Styles (or combinations of styles) used were: Italian time in use till 1789, in which the hours were computed starting from twilight; apparent solar time based on the actual motion of the sun; mean solar time based on the average motion of the sun; local time referred to the actual passage of the sun across the local meridian (local culmination); French time starting at midnight and regulated on the local culmination; Western European Time regulated on the culmination of fictitious average solar motion on a reference meridian 15° East. A test was performed to verify whether the times chosen for readings were appropriate, in particular when observations were performed not close to the daily minimum and maximum. In effect, in the early period with Morgagni and Toaldo, the choice of schedule of observations was good, but afterwards the introduction of new observations, not always established at the most appropriate schedule, reduced the representativity of the data. The errorin calculating the daily average temperature after a given number of observations taken at different hours of the day has been analysed. National, and especially international recommendations have been particularly important in the choice of observations times, and in determining averages. These recommendations have been simultaneously applied on a large number of sites, causing an in-homogeneity that may be misinterpreted as a well-documented, widespread climate change.