Spatiotemporal analysis of nonlinear trends in precipitation over Germany during 1951–2013 from multiple observation‐based gridded products

Abstract

Spatial and temporal patterns of trends in annual and seasonal precipitation over Germany during 1951-2013 were analysed using the ensemble empirical mode decomposition (EEMD) method. Three widely used and recognized high-resolution observation-based gridded precipitation products, the Climatic Research Unit (CRU) time-series data, Global Precipitation Climatology Centre (GPCC) Full Data Reanalysis data and the EU-FP6 project ENSEMBLES derived data set (EOBS), were used and compared. Comparison among different products showed that the CRU used a considerably lower number of gauge stations (maximum 36) in the construction of gridded product over Germany and should be used with great caution for other applications, and overall CRU presented less precipitation and smaller inter-annual variability; GPCC and EOBS agreed well in the spatial and temporal distribution of seasonal and annual precipitation, as well as the identified trends. The spatiotemporal analysis of trends showed that trends in precipitation during 1951-2013 were, in most cases, highly nonlinear and varying over time. In particular, the spring, summer and autumn precipitation showed large variations in trends. Therefore, the assumption of linear trends on which previous studies were based is invalid. Annual precipitation showed increasing trends in over 66% of Germany (particularly northern and eastern Germany), and the magnitude of increasing trends were generally enhanced over time, e.g. GPCC showed that the average magnitude of increasing trends rose from 6.3% for the 1951-1980 period to 12% for the 1951-2013 period. There were considerable spatial and temporal variabilities in trends in seasonal precipitation totals over Germany. This study is among one of the first studies applying the EEMD method to a comprehensive analysis of time-varying trends in precipitation over Germany, which is expected to improve our understanding of the complex and nonlinear regional climate system.

Observed changes in precipitation during recent warming: The Czech Republic, 1961–2019

“…Duan et al . (2019) analysed nonlinear trends in annual and seasonal precipitation series in Germany for the 1951–2013 period, using three different gridded datasets. Annual precipitation trends were positive over 66% of the territory, particularly in northern and eastern Germany, and their magnitude generally intensified over time.…”

“…Precipitation analyses on a scale that covers all of Europe have focussed on long‐term precipitation trends (e.g., Caloiero et al ., 2018), precipitation extremes (e.g., Klein Tank and Können, 2003; Moberg et al ., 2006; Zolina, 2012; van den Besselaar et al ., 2013; Łupikasza, 2017; Deumlich and Gericke, 2020) and model projections for them (e.g., Frei et al ., 2006; Nissen and Ulbrich, 2017; Rajczak and Schär, 2017). Many publications address the homogenization and preparation of datasets and related studies on national or regional scales (e.g., Auer et al ., 2005; Zolina et al ., 2008; Niedźwiedź et al ., 2009; Łupikasza et al ., 2010; Zahradníček et al ., 2014; Gajić‐Čapka et al ., 2015; de Leeuw et al ., 2016; Markonis et al ., 2017; Milovanović et al ., 2017; Jaagus et al ., 2018; Murphy et al ., 2018; Duan et al ., 2019; Kocsis et al ., 2020; Twardosz and Cebulska, 2020).…”

Re-evaluation of the Power of the Mann-Kendall Test for Detecting Monotonic Trends in Hydrometeorological Time Series

“…Compared to parametric tests (e.g., regression coefficient test), non-parametric tests (e.g., the MK test and Spearman' rho test) have no requirements of homoscedasticity or prior assumptions on the distribution of the data sample (Önöz and Bayazit, 2003) and are less sensitive to outliers (Hamed and Ramachandra Rao, 1998;Hamed, 2007). As the MK test statistic is determined by the ranks and sequences of time series rather than the original values, it is robust when dealing with nonnormally distributed data, censored data, and time series with missing values (Hirsch and Slack, 1984), which are commonly encountered in hydrometeorological time series (Duan et al, 2018(Duan et al, , 2019Gao et al, 2018Gao et al, , 2019Dong et al, 2019).…”

Trend Analysis of Rainfall Time Series in Shanxi Province, Northern China (1957–2019)

“…In recent years, many studies related to the trends and variability in precipitation under global warming have investigated different regions, such as India [4,5,[9][10][11][12][13], the Mediterranean [14][15][16][17][18], Ethiopia [19], China [1,2,[20][21][22][23][24][25], the Netherlands [26], South Korea [27], Brazil [28], Germany [29], and West Africa [30]. Although various studies on rainfall variation and trend analysis have been investigated in western [31,32], eastern [24], southern [33], north-west [34], south-east [2], and south-west China [23], little investigation has been carried out in the northern regions (e.g., Shanxi province, China).…”