This chapter assesses the scientific literature describing expectations for near-term climate (present through mid-century). Unless otherwise stated, ‘near-term’ change and the projected changes below are for the period 2016–2035 relative to the reference period 1986–2005. Atmospheric composition (apart from CO2; see Chapter 12) and air quality projections through to 2100 are also assessed.

Decadal Prediction

The nonlinear and chaotic nature of the climate system imposes natual limits on the extent to which skilful predictions of climate statistics may be made. Model-based ‘predictability’ studies, which probe these limits and investigate the physical mechanisms involved, support the potential for the skilful prediction of annual to decadal average temperature and, to a lesser extent precipitation.

Predictions for averages of temperature, over large regions of the planet and for the global mean, exhibit positive skill when verified against observations for forecast periods up to ten years (high confidence). Predictions of precipitation over some land areas also exhibit positive skill. Decadal prediction is a new endeavour in climate science. The level of quality for climate predictions of annual to decadal average quantities is assessed from the past performance of initialized predictions and non-initialized simulations. {11.2.3, Figures 11.3 and 11.4}

In current results, observation-based initialization is the dominant contributor to the skill of predictions of annual mean temperature for the first few years and to the skill of predictions of the global mean surface temperature and the temperature over the North Atlantic, regions of the South Pacific and the tropical Indian Ocean for longer periods (high confidence).