Abstract. For sectors like agriculture, hydrology and ecology, increasing interannual
variability (IAV) can have larger impacts than changes in the mean state,
whereas decreasing IAV in winter implies that the coldest seasons warm more
than the mean. IAV is difficult to reliably quantify in single realizations
of climate (observations and single-model realizations) as they are too
short, and represent a combination of external forcing and IAV. Single-model initial-condition large ensembles (SMILEs) are powerful tools to overcome
this problem, as they provide many realizations of past and future climate
and thus a larger sample size to robustly evaluate and quantify changes in
IAV. We use three SMILE-based regional climate models (CanESM-CRCM,
ECEARTH-RACMO and CESM-CCLM) to investigate downscaled changes in IAV of
summer and winter temperature and precipitation, the number of heat waves, and
the maximum length of dry periods over Europe. An evaluation against the
observational data set E-OBS reveals that all models reproduce observational
IAV reasonably well, although both under- and overestimation of
observational IAV occur in all models in a few cases. We further demonstrate
that SMILEs are essential to robustly quantify changes in IAV since some
individual realizations show significant IAV changes, whereas others do not.
Thus, a large sample size, i.e., information from all members of SMILEs,
is needed to robustly quantify the significance of IAV changes. Projected
IAV changes in temperature over Europe are in line with existing literature:
increasing variability in summer and stable to decreasing variability in
winter. Here, we further show that summer and winter precipitation, as well
as the two summer extreme indicators mostly also show these seasonal
changes.