An ensemble is a group of climate model simulations used for climate projections. Rather than running a single climate model, an ensemble of thousands of model versions is run with each version of the model being slightly different from one another. This produces a result with multiple scenarios. Climate models can help understand how the Earth’s climate works. Climate models are constrained by observations in the real world and can help inform further observational efforts. They are based on physical principles such as physics, chemistry, and biology, on how the Earth works.They can help predict future climate projections using past and present day information, such as modelling global temperature change. These models can help explain the ‘what ifs’ by running multiple scenarios at once. An ensemble of models can help represent new resources for studying ranges of plausible climate change responses in relation to a given forcing. Forcing means the amount of energy that the Earth receives from the sun and the amount we radiate back to space. These forcing variances are determined by factors that influence our atmosphere such as greenhouse gases. Using an ensemble approach is quite beneficial. When comparing against gridded data (two-dimensional data), ensemble results have come closest to replicating historical climate projections. Ensemble projections which accurately project historical data are more likely to represent future climate scenarios more efficiently.
This type of ensemble requires varying the initial conditions of all of the models, such as their starting temperatures, winds and humidity. Since the climate system is chaotic, small changes in these conditions can lead to drastically different paths for the system as a whole.
Ensembles made with the same model but different initial conditions only characterize the uncertainty associated with internal climate variability, whereas multi-model ensembles including simulations by several models also include the impact of model differences. Therefore these ensembles can be a good method to validate the type of model used, by comparing it to the real-world climate system's past development, known as a hindcast (in contrast to a forecast).
In climate models, certain parameters carry a larger uncertainty than others, thus making it hard to determine the true effect they have on the final results. In perturbed parameter ensembles, model parameters are varied in a systematic manner which aims to produce a more objective estimate of modelling uncertainty. The most thorough way to do this is to run a massive ensemble experiment which test all relevant parameters in every possible combination with one another.
A model can be introduced to different types of forcing, so a forcing ensemble is designed to represent uncertainty in past and future forcing of climate change projections.
A grand ensemble is a minimum of two or more nested ensembles.