Climate change scenarios are generated using climate models along with assumptions with regard to future greenhouse gas emissions. It is important to use models with a high resolution for Switzerland, given its complex topography. Even then, it is not possible to reliably and fully map processes occurring on a small scale, and this can lead to uncertainties. Uncertainties arise from the fact that natural fluctuations can be superimposed on the changes brought about by greenhouse gases.
Background Information on Climate Change Scenarios
How the climate of Switzerland will look in the future depends on global trends on the one hand, while on the other, regional and local factors play an important role. Climate model simulations are employed to enable climatologists to take into account the numerous influencing factors when generating regional climate scenarios. Pre-defined evolution paths for greenhouse gas emissions (emission scenarios) and their levels of concentration in the atmosphere constitute one of the framework conditions for projections of future climate scenarios.
Projections using climate models
Climate models are primarily based on the same mathematical physics equations that underlie weather forecasts. In addition to these, however, the models also factor in other important components of the climate system, such as oceans, sea ice and land surfaces, coupling these with atmospheric factors. The atmospheric conditions for the next time segment are determined by solving the atmospheric equations at the individual grid points within a grid network that, in the case of a global model, spans the entire planet. For such global models, time segments of between 10 and 30 minutes are the norm. This means that for a simulation covering a period of 100 years, equations for around two million time segments must be calculated.
Regional models with higher resolution
The resolution of a climate model is defined by the distance between the individual grid points. For global-scale models, this distance is generally between 100 and 300 kilometres, owing to limits in computational capacity. In the case of areas with complex topography such as Switzerland, this is too low a resolution to be able to produce an accurate representation of localised processes. This is where regional climate models come into play. Although such models are only applied to a small spatial area (e.g. Europe), they run with a much higher resolution - typically 10 to 50 km. On the peripheries of the simulation area, data output from global-scale models are used to drive the regional climate models.
In spite of the relatively high resolution, it is not possible to compute small-scale effects, such as the impact of mountain topography on wind flow, local exchange processes between the ground and the atmosphere, or the local effects of clouds on irradiation. These influences have to be described more simply, based on empirical data. The nature of this description varies from model to model, and is one of the main reasons why different models generate different climate projections. To take account of this kind of uncertainty, several different regional and global-scale climate models are normally evaluated together.
Climate changes can also occur as a result of natural phenomena. Particularly where a small region like Switzerland is concerned, and for timescales that extend into the near future (e.g. the first half of the 21st century), natural long-term climatic fluctuations play a major role. Over the course of several decades, such variability can be superimposed over long-term trends - for instance those caused by changes in greenhouse gas emissions. These natural fluctuations can be triggered by gradual variations in the surface temperature of the Atlantic ocean and corresponding changes in atmospheric circulation over the European continent. The Swiss climate change scenarios account specifically for the uncertainties that such natural fluctuations give rise to.
The temperature and precipitation changes for the various emission scenarios are fraught with uncertainties. The coloured bars in the charts represent numerous change values. The exact position of the bars is determined by means of a statistical analysis. However, for a number of reasons, the reliability of this uncertainty estimate is still considered to be relatively poor. It is therefore essential, for example, to make some basic assumptions with regard to model biases. Thus, probabilities cannot be assigned to the uncertainty bars. Instead, three plausible change values are derived for each uncertainty bar: a low estimate (bottom end of the bar), mean estimate (black line) and upper estimate (upper end of the bar).