Scientists tend to apply very specific meanings to words that are more freely used in everyday life. And confusingly, terms like prediction may be used differently in different scientific communities. In the study of climate, meteorology and oceanography, prediction is used mostly in the same way as forecast.
A prediction is the most probable outcome of future developments – in terms of a number, an interval or a category. A prediction is constructed using statistical or dynamical models, which process the knowledge given by an initial state or given by predictions of major steering components.
Another approach to envisage futures is to construct scenarios. These are descriptions of alternative possible, plausible, internally consistent, but not necessarily equally probable futures. In climatology, this is done by simulating in climate models the implications of factors such as changing land use and changing atmospheric compositions (increases in greenhouse gases). Thus scenarios are, technically, conditional predictions – i.e., conditional on the assumptions made about the future evolution of greenhouse gases and so on.
Climate changes over the time frame of several decades up to a few centuries are explored using scenarios. These futures are also referred to as projections of climate change. The term projection is mostly used in the same way as scenario. Widely-used climate scenarios/projections are based on the IPCC SRES scenarios of socio-economic developments and emissions.
General seasonal forecasts are now experimentally prepared by many weather services. Another experimental avenue examines the potential of the present state of slow climate components as initial states for predictions across one or two decades, in particular the ocean. For longer lead times than a few centuries, such as thousands of years, predictions of natural climate change again become possible. This is because of the dominance of an easily predictable steering component, namely the characteristics of the Earth’s orbit. But to understand how natural and anthropogenic forcing may interact on these long timescales requires long-term emissions scenarios.