Influence of data uncertainty on cold season threshold-based climate indices

• Published in: Meteorologische Zeitschrift (Berlin, Germany : 1992) Vol. 32; no. 3; pp. 195 - 206
• Main Authors: Bell, Louisa Marie, Schlünzen, K. Heinke, Sieck, Kevin
• Format: Journal Article
• Language: English
• Published: United States Schweizerbart Science Publishers 11.09.2023; Borntraeger
• Subjects: climate index; data uncertainty; e-obs; euro-cordex; Meteorology & Atmospheric Sciences; threshold; winter
• ISSN: 0941-2948
• DOI: 10.1127/metz/2023/1158

Climate indices are used to reduce the complex climate system and its changes to simple measures. The data basis – whether observational data or climate model data – to which the climate indices are applied, is usually subject to uncertainties. For threshold-based climate indices, the data uncertainty influences the threshold value, and, hence, the uncertainty can influence the values for the climate index. What the actual impacts of these uncertainties are on threshold-based climate indices is examined in this paper. The focus is not only on the climate model uncertainty, but also on the observational data uncertainty. The general sensitivity of each of the chosen climate indices to arbitrary changes in the threshold is studied. This shows a higher sensitivity of indices assessing extremes (ice days, heavy precipitation days) to changes in the threshold than indices that integrate a quantity over a given time interval (coldsum, consecutive days). For assessing an ensemble of climate model data with respect to their ability to reproduce the index values for current climate, the reference data uncertainty is applied to the chosen threshold-based climate indices by changing their threshold value by its corresponding uncertainty. It is shown that the climate model uncertainty can be within the range of the reference data uncertainty. When using threshold-based climate indices to assess changes in future climate periods, uncertainties should always be taken into account and ideally corrected in an appropriate way. This is especially important for indices that assess extremes