Why can we make robust statements about possible future climate change, although our weather forecast for more than 10 days is pretty shaky? Well – weather is not climate. Weather is defined as the actual state of the atmosphere. Climate, however, is the statistical behavior of weather, or in short, weather statistics. We cannot predict the motion of single air molecules, we can, however, predict the temperature change, i.e. the statistics of many air molecules, if we heat the air. Similar considerations apply to weather and climate.
Modern climate research uses a wider definition of climate. To understand how the long-term state of the atmosphere changes with time, we have to consider the interaction of the atmosphere with the other components of the climate system (see figure). Models of the climate system compute, over many years and decades, the motion of the atmosphere, i.e., the statistics of weather, die ocean currents, the drifting of sea ice, the expansion and retreat of ice sheets (both are components of the cryosphere), the heat and water storage in the soils (the pedosphere), the migration of vegetation. With this model system we do experiments to understand past, present and possible future climate changes. In addition to modelling, we also work with measurement of climate variables, e.g. with satellite sensors.