Trends in extreme temperature indices in Austria based on a new homogenised dataset



Instrumental time series are often affected by inhomogeneities which can mask or amplify climate change signals. Various procedures for the detection and adjustment of breaks exist for monthly and annual time series. Homogenization methods on a daily basis are scarce and often disregard uncertainties accompanying the break adjustment. We present a complete homogenization procedure for daily extreme temperature series based on the break detection method PRODIGE and SPLIDHOM for break correction. Both parts of the homogenization rely on the existence of highly correlated reference stations. After the statistical comparison with neighbouring stations, detected breaks are verified and further localized by metadata. Uncertainties of the break adjustments are estimated by altering reference stations and by applying a bootstrapping technique providing an objective indication about the reliability of the homogenization.

The method was tested and applied to 71 time series of daily minimum (TN) and maximum temperatures (TX) in Austria covering the period 1948–2009. For some series homogenization was not possible due to large uncertainties in the adjustments or a lack of suitable reference series. In the remaining 57 TN and 54 TX series a total number of 139 breaks were detected. Seventy-five percent of those breaks are documented in the metadata archive, with most of them being caused by station relocations and instrumentation changes. In general, the mean over the temperature dependent adjustments of all stations show a temperature reduction. However, the majority of breaks have mean amplitudes of less than 0.5 °C. A comprehensive analysis was performed on the new homogenized daily dataset, showing a widespread warming trend in both TN and TX series. The warming trend is in general amplified due to the homogenization. However, significant changes in the trend are only observed at very few stations. In autumn, however, the trend is reversed in many temperature based ‘climate change detection indices’. Copyright © 2012 Royal Meteorological Society