Jones, Roger ORCID: 0000-0001-6970-2797 and Ricketts, James (2015) Analysing steps in modelled global surface air temperature. Working Paper. VISES, Victoria University, Melbourne, Australia.

Abstract

Here, we apply a multi-step bivariate test to detect shifts in modelled global surface air temperature to see whether historical simulations match the pattern of shifts in observations, assess so‐called ‘hiatus’ periods between shifts and investigate the evolution of steps and trends under different emissions pathways. Key findings include: simulations reproduce the broad pattern of historical steps reasonably well, the 1996–2005 decade including the large 1996–98 step change, and total warming to 2005 shows no relationship with model equilibrium climate sensitivity (ECS). Nor does 20th century warming show any relationship with 21st century warming. Fifty‐five percent of models produce a step change in 1996–98, and are followed by ‘hiatus’ periods of 7–26 years, eight being equal to or greater than the current 18 years. For the representation concentration pathway RCP4.6 multi‐model ensemble (MME), total step changes 2006–95 are 3.5 times as effective in explaining total warming and climate sensitivity than internal trends. Under the lower RCP2.6 pathway, the MME stabilises around 2050 and the timing of stabilisation is dominated by climate variability rather than ECS. The finding that warming in both models and observations is dominated by shifts shows that the current emphasis on trend analysis alone is inadequate for analysing climate change. This has substantial implications for how climate risk is framed, analysed and communicated.

Search Google Scholar

Repository staff login