Clouds over Europe¶
Cloud formation and dissipation processes are still not well understood due to the complex interaction of many physical mechanisms. At the same time, clouds play an essential part in the global climate system. However, since their behavior is still not well understood, the greatest uncertainties in current climate projections are in this area. To gain a better understanding of cloud-forming processes and to study their spatiotemporal distribution, data from geostationary meteorological satellites, such as the Meteosat series of EUMETSAT, can be used. So let’s take a look at the cloud distribution over Europe from the perspective of the Meteosat Second Generation satellites (MSG).
Below you can see the VIS and IR composites of an exemplary MSG scene:
MSG satellites scan the Earth’s atmosphere every 15 minutes in 12 spectral bands (from VIS to thermal infrared):
Research objective¶
In this case study we want to use MSG data from 2005 and 2006 to investigate the average cloud distribution and cloud altitudes over Europe during this period. Our goal is to develop two ML models:
One which we can use to generate a spatially explicit cloud product for every MSG scene available
and another one to analyze cloud altitudes.
Due to data size issues, we will only use data from the 15th of each month at a three-hourly frequency. This ensures, that diurnal and seasonal fluctuations are still covered while reducing data size significantly.
For training and validation purposes, we’ll use METAR station data gathered at 273 European airports. At these stations, among many other variables, cloud cover and cloud altitude are recorded.
Additionally, a digital elevation model is provided:
Task¶
Develop a conceptual approach for this task and discuss it in your group. Create a flowchart of all the steps necessary to implement your approach.
Ask yourself:
Are the data sets provided suitable to reach the goal (spatially explicit cloud product)?
Which problems might occur and how could you address these?