Robust changes to the wettest and driest days of the year are hidden within annual rainfall projections – a New Zealand case study

Abstract
Understanding how the statistical properties of daily rainfall will respond to a warming climate
requires ensembles of climate model data which are much larger than those typically available from
existing centennial-scale modelling experiments. While such centennial-scale experiments are very
useful to explore scenario uncertainty in twenty-first century climate, ensemble size constraints
often result in regional climate change assessments restricting their focus to annual- or
season-mean rainfall projections without providing robust information about changes to the most
extreme events. Here, we make use of multi-thousand member ensembles of regional climate
model output from the Weather@Home project to explicitly resolve how the wettest and driest days
of the year over New Zealand will respond to simulations of a 3 ◦C world, relative to simulations of
the climate of the recent past (2006–15). Using a novel framework to disentangle changes during
the wettest and driest days of the year, we show that many regions which show negligible change in
annual mean rainfall are in fact experiencing significant changes in the amount of rain falling
during both the wettest and driest spells. Exploring these changes through the lens of drought risk,
we find many agricultural regions in New Zealand will face significant changes in the frequency of
low-rainfall extremes in a warmer world.