Little evidence of hysteresis in regional precipitation, when indexed by global temperature rise and fall in an overshoot climate simulation

Society has set the aim of stabilising climate at key temperature thresholds, such as global warming at or below 1.5°C or 2.0°C above preindustrial levels. However, greenhouse gas emissions are failing to decline, and if they continue on their current trajectory it is likely that such thresholds will be crossed in the decades ahead. Because of this risk, there is an emerging focus on overshoot, where, for a temporary period, global warming is allowed to cross critical thresholds to reach a peak value before decreasing to the desired limit. A key question about overshoots is whether there are hysteresis effects—that is, whether global or regional climate has properties that differ between the phase of global warming increase and the phase of decreasing. Here, we analyse temperature and precipitation data from five Earth System Models (ESMs) forced by the SSP5-3.4-OS CMIP6 overshoot scenario. We look at the level of precipitation during two periods of near-identical global warming: one whilst temperatures are rising, and the other when they are falling. For global means, we find a statistically significant difference between precipitation values during the two periods. This is an example of hysteresis, as the reversion to an earlier global warming state results in a level of global rainfall which is different from that observed when warming was increasing. Spatial disaggregation of rainfall differences between the two near-identical warming levels shows the largest differences in the tropical region, which are statistically significant for four of the five ESMs. When considering much smaller regions, including parts of the tropics, there remains some evidence of hysteresis. However, the differences are no longer statistically significant against a background of substantial interannual rainfall variability. We discuss the implications of our findings for climate impacts assessments.