Climate variability conceals emerging hydrological trends across Great Britain

Abstract

Detecting a climate change signal from observed trends in river flows and hydrological extremes is challenging given the limited length of observations and the effects of internal climate variability. There has been an increasing call to better integrate historical observations with model projections, particularly given apparent inconsistencies between observed and projected hydroclimate trends. Here we use the UK as a case study of a region with apparent incongruity between past trends and future projections, such as observed summer wetting but broad agreement between climate models of reduced summer rainfall and river flows. Applying dynamical adjustment shows empirically that internal atmospheric circulation variability was a dominant factor in the observed positive summer rainfall trends over 1981–2010. Characterising the impacts of internal climate variability is crucial to fully appraising the range of possible hydrological extremes in current and future climate. Hence, we use a single model initial condition large ensemble (SMILE), with RCP8.5 forcing, to drive hydrological models at 190 catchments to explore the wide range of past and future river flow and hydrological drought trends that could arise due to internal variability. The results place the observed trends in context, showing that large ensembles are needed to fully capture the range of variability. This includes robust drying and wetting trends that could have occurred, thus in part reconciling the fact that observed trends may at first seem inconsistent with projections. Our results further show that the timing of a robust climate change signal above historical variability (i.e., a Time of Emergence) in river flows may remain obscured for decades due to the range of hydrological variability. There are however clear hotspots, such as decreasing low flows in southwest England, with an imminent ToE. However, a late ToE does not negate the potential for increased risk and adaptation measures should be formulated before a statistically significant climate signal emerges.