Reconciling observed and modeled temperature and precipitation trends over Europe by adjusting for circulation variability

Abstract

Europe experienced a pronounced winter cooling of about −0.37°C/decade in the period 1989–2012, in contrast to the strong warming simulated by the Coupled Model Intercomparison Project Phase 5 multimodel average during the same period. Even more pronounced discrepancies between observed and simulated short‐term trends are found at the local scale, e.g., a strong winter cooling over Switzerland and a pronounced reduction in precipitation along the coast of Norway. We show that monthly sea level pressure variability accounts for much of the short‐term variations of temperature over most of the domain and of precipitation in certain regions. Removing the effect of atmospheric circulation through a regression approach reconciles the observed temperature trends over Europe and Switzerland and the precipitation trend along the coast of Norway with the corresponding multimodel mean trends.

Temperature monitoring in mountain regions using reanalyses: lessons from the Alps

“…MESCAN-SURFEX shows particularly large errors with a clear negative bias from the 1960s to about 2000 with errors up to −3 K in individual years. Also HARMONIE tends to show predominantly negative biases between 0 and −1 K. Interestingly, all reanalysis data sets strongly underestimate the temperature anomalies in the years 1989 to 1993 by about 1 to more than 2.5 K. This is the period right after the abrupt warming around the year 1988 (Reid et al 2016, Saffioti et al 2016, Sippel et al 2019. E-OBS HOM, which in general is closest to the SM benchmark, does not show any bias peculiarities from 1989 to 1993 ( figure 3, right panel).…”

Effect of atmospheric circulation on surface air temperature trends in years 1979–2018

“…Similarly, the circulation adjustment in R19 greatly smoothed the seasonal cycle of monthly mean temperature trends in Finland in the years 1979-2018. Furthermore, Wallace et al (2012), Deser et al (2016), Saffioti et al (2016) and R19 all found the circulation-adjusted temperature trends to agree better with multi or single model ensemble mean temperature trends in climate models than the original observed trends did.…”

“…Parker (2009 also found a positive albeit small (0.09 °C) circulation-induced anomaly in the annual mean of the Central England Temperature in the last decade of his study (1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005). Conversely, Saffioti et al (2016) reported a circulation-related winter cooling of nearly 0.8 °C per decade in Europe in the years 1989-2012. Some of these differences may be method-related, but as argued later in Sect.…”

“…Although such internal variability is largely smoothed out in averages over several tens of model simulations, it both complicates the interpretation of observed climate changes and is an important uncertainty in projections of regional climate change in the next several decades (Deser et al 2012(Deser et al , 2014(Deser et al , 2016Sutton 2009, 2010). A major fraction of internal variability of surface climate results from variations in atmospheric circulation (Parker 2009;Deser et al 2012Deser et al , 2014Saffioti et al 2016Saffioti et al , 2017. Specifically, surface air temperature is affected by atmospheric circulation both via advection and because circulation modulates diabatic heating, for example by regulating cloudiness (Holmes et al 2016;Räisänen 2019a).…”

“…Many of them have related the variations of temperature to the distribution of monthly mean sea level pressure (SLP), using a variety of statistical techniques. For example, Wallace et al (2012) and Smoliak et al (2015) applied partial linear regression, Deser et al (2016) constructed analogues, and Saffioti et al (2016) linear regression using the principal components of SLP as the predictors. In an alternative, more data intensive approach, Parker (2009) used two-dimensional and Räisänen (2019b, hereafter R19) three-dimensional trajectories of air calculated from 6-hourly pressure or wind data.…”