Some definitions of our seasonal, decadal and centennial climate predictions.
Although climate is expected to warm over the coming century in response to increasing levels of greenhouse gases, regional changes over the coming seasons to a decade or more are likely to be dominated by unforced natural variability of the climate system. Some of this natural variability is potentially predictable months or even years in advance because it is related to relatively slow processes in the ocean, such as El Niño, fluctuations in the thermohaline circulation, and large-scale anomalies of ocean heat content. Seasonal to decadal forecasts therefore attempt to predict natural variability in addition to externally forced changes. This is achieved by starting the predictions from the current observed state of the climate system, as well as specifying changes in anthropogenic sources of greenhouse gases and aerosol concentrations and projected changes in solar irradiance and volcanic aerosol. However, there are sometimes large uncertainties in these forecasts because the observations with which the models are initialised are incomplete, and the models themselves are imperfect. These uncertainties are taken into account by making an ensemble of forecasts, intended to cover the range of possible outcomes consistent with imperfections in the initialisation and modelling systems. In addition to estimating uncertainties these ensembles can also be used to provide probabilistic information that can be extremely valuable in many areas of industry and commerce.
The latest seasonal forecasting system, GloSea4, uses a member of the HadGEM3 family of models which is currently under development. The current decadal forecasting system, DePreSys, uses the HadCM3 model. It is planned that climate configurations of the MetUM at all timescales will move to the HadGEM3 family over the next few years.
Centennial climate forecasts involve running a long (~1000 years) control run representing the climate in pre-industrial times (~1860), from which other "scenario" runs are initialised. These may include some or all of the known climate forcings over the 20th century, such as natural and anthropogenic sulphur emissions, atmospheric aerosols such as sea salt, dust, volcanic emissions, ozone, greenhouse gases (e.g. carbon dioxide, methane, CFCs). Scenarios of how such forcings may change in the future are then applied as a means of exploring how human activities may change the composition of the atmosphere, how this may affect global climate, and how the resulting climate changes may impact upon the environment and human activities. This provides a set of internally-consistent pictures of possible future climates, each dependent on a set of prior assumptions.
Earth System configurations include the ecosystem and chemistry components of the climate system in order to account for feedbacks in the climate system not traditionally included in climate models, such as those involving the terrestrial and oceanic ecosystems or atmospheric chemistry. These have generally been excluded from climate models on the grounds of computational cost, but increases in computing power now make inclusion of these components viable. The Met Office will provide centennial climate predictions from an Earth System model to the next IPCC Assessment Report.
The regional climate configuration provides a high-resolution representation of a limited region and can be located anywhere in the world. Due to their higher resolution, regional models provide a more detailed description of orographic effects, land-sea contrast and land-surface characteristics, and also give an improved treatment of fine scale physical and dynamical processes. Regions typically include areas of research interest such as Europe, South and East Asia and Africa. The limited-area regional model is embedded in a representation of the observed global climate derived from the weather observations used as initial conditions for weather forecasting models. It is also embedded in representations of the climate derived from global climate models to add realistic detail to global climate simulations or climate change projections. In both cases, the atmosphere component is run in standalone mode using observed or modelled time-varying sea surface temperatures and sea ice distributions.