The rate of global surface warming is a crucial observable quantity for tracking progress towards global climate targets. Its near-term evolution is however also strongly influenced by interannual-to-decadal variability, which can hamper detection of the effects of emission mitigation. Hence, process-based approaches are needed that can reduce this variability, by separating interannual fluctuations from forced and longer-term changes.

We present a new such approach, based on Green’s functions that relate fluctuations in global mean surface air temperature (GSAT) to the monthly geographical pattern of sea-surface temperatures. For each month, a contribution to GSAT from internal variability is calculated, and subtracted from the total as a physics based filtering of the total GSAT anomaly relative to 1850-1900.

For near-term warming rates under differing assumptions, we show that our approach can advance separation between the climate responses to low and high emission scenarios by up to a decade.

Our filtering approach reduces the diagnosed rate of surface warming over the most recent decade (2011-2020), which was influenced by the El Nino of 2015-2016, from the observed 0.35 °C per decade in the HadCRUT5 dataset, to 0.24 °C per decade, consistent with the 50-year trend (1971-2020) from the same dataset. Conversely, the rate over the so-called “global warming hiatus” period (2001-2010), which was observed to be only 0.08 °C per decade, strengthens to 0.21 °C per decade using our method.

We suggest that such filtered warming rates could represent a strong addition to the tools used by the climate community to inform policy makers and stakeholder communities, provided an effort is made to develop, improve and validate standardized Green’s functions.