The synergy of new instrumentation for observing the atmosphere with modeling of the atmosphere from physical principles and with evaluating data by machine learning and artificial intelligence (AI) is a key strength of ICE-4. For example, global circulation models are an important tool to gain understanding of the complex and chaotic processes within our atmosphere. These models contain always approximations, which require tuning by observational data. Novel predictive models based fully on machine learning require even more observational data to become reliable. Other numerical models are designed to simulate very specific physics within the atmosphere such as emission and absorption of light, the formation of clouds or the propagation of atmospheric waves. To further our understanding of these processes, novel instruments measuring the atmosphere are needed that provide the required data from aircraft, balloons, or satellites. Measurement platforms such as zero-pressure balloons going up to 35 km altitude or the German research aircraft HALO allow our instruments to have a close look at one of the last frontiers that are surprisingly unknown. Today, novel observations still uncover new atmospheric phenomena and chemical reactions that are missing in modelling efforts. Such new observations then in stigate a flurry of activities in the scientific community that often result in better forecasts of weather, boundary layer pollution, or extreme weather events.
For example, one still open question is the general speed of atmospheric large-scale turn-over. Due to its slowness, it is very difficult to measure this circulation directly. However, the stratospheric circulation has profound impact on long-term climate forecasts. We use stratospheric balloons carrying our GLORIA and air-core instruments, or satellite-borne remote sensing instruments to measure highly accurate trace gas concentrations of long-living anthropogenic substances. Combining these unique measurements with the predictions of our global-circulation model CLaMS for different configurations, one can derive estimates about both the current speed of the circulation as well as determine whether or not this circulation system is in the process of change.