Accepted Article Homogenization of daily ECA&D temperature series Antonello Angelo Squintu, Gerard van der Schrier, Yuri Brugnara, Albert Klein Tank September 17, 2018 abstract The daily maximum and minimum temperature series of the European Climate Assess- ment & Dataset are homogenised using the quantile matching approach. As the dataset is large and the detail of metadata is generally missing, an automated method locates breaks in the se- ries based on a comparison with surrounding series and applies adjustments which are estimated using homogeneous segments of surrounding series as reference. A total of 6500 series have been processed and after removing duplicates and short series, about 2100 series have been have been adjusted. Finally, the effect of the homogenization of daily maximum and minimum temperature on trend estimation is shown to produce a much more spatially homogeneous and then plausible picture. 1 Introduction Modifications to meteorological stations, such as relocation, replacement of the instrument, re- calibration, new buildings in the neighborhood or growth of vegetation in the proximity, alter temperature measurements and introduce biases in the observational records that do not relate to weather and climate [Aguilar et al., 2003, Hartmann et al., 2013]. The analysis of climatic vari- ability and climatic change requires homogeneous temperature series [Peterson et al., 1998]: these series do not confuse the climatic signal with artificial biases which are present in non-homogenous series [Menne and Williams Jr, 2009, Brunetti et al., 2006, Thorne et al., 2005, Begert et al., 2005]. Prior to climate analyses, actions are required aimed at the removal of step-like or gradual changes related to these non-climatic effects in observational records [Caussinus and Mestre, 2004]. The registration of activities on meteorological stations in the metadata keeps track of these changes and sufficiently detailed metadata allow a precise temporal localization of the breaks. Unfortu- nately the availability of metadata is often low, especially further back in time, and doesn’t cover the whole set of inhomogenities that affect the measurements [Caussinus and Mestre, 2004]. This implies that break-detection based on metadata only is not possible for many datasets, even though this approach is regarded as most accurate and reliable. This argument, and the sheer size of a dataset, motivates the use of an automated homogenization procedure [Caussinus and Mestre, 2004]. The aim of this study is to develop a pan-European homogeneous dataset of daily maximum and minimum temperature using such an automated homogenization procedure. It will use a recent agreement-based system to detect breaks [Kuglitsch et al., 2012] and the quantile matching technique [Trewin, 2013] in combination with a pairwise-comparison [Menne and Williams Jr, 2005] approach to determine adjustments. The elements in this approach are introduced below. Automated homogenization procedures consists of two steps: break detection and adjustment calculation (which follow - or are integrated with - a quality check procedure) [Alexandersson, 1986, Caussinus and Mestre, 2004]. These have been focusing mainly on the detection of breaks in the monthly, seasonal or annual values and use statistical tests accompanied with penalizing functions [Alexandersson, 1986, Caussinus and Mestre, 2004, Menne and Williams Jr, 2005, Wang et al., 2007] or inspections on autocorrelation of residuals [Vincent, 1998]. Recent comparisons [Venema et al., 2013,Domonkos, 2013,Lindau and Venema, 2013] have pointed out advantages and drawbacks of the most common systems. Procedures that look for an agreement among methods (e.g. [Kuglitsch et al., 2012]) go one step further and take benefits from the reduced uncertainty in break location by looking for consensus. Homogenization of annual or monthly averages does not automatically imply a homogenization of higher-order moments [Trewin, 2013] since the processes that generates inhomogeneities on daily This article is protected by copyright. All rights reserved. This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1002/joc.5874 source: https://doi.org/10.7892/boris.120719 | downloaded: 10.6.2020