How important is airplane data for weather forecasting?
The German weather service just stopped using it.
Airplane data is a large contributor to international observation networks. However. the German weather service just stopped using airplane measurements.
In combination with weather-stations, satellite, radar and radiosonde observations, national weather services run “assimilations” as initial conditions for weather models.
Data assimilations for weather-models try to calculate the state of the entire atmosphere at “hour 0” for weather-models. This is challenging because only a tiny fraction of the world is covered with measurement stations and they tend to be fairly close to ground. Without data, the current weather conditions for large portions of the world, have to be calculated from nearest observations. For areas like the Atlantic Ocean, there are not many observations available.
Satellites provide additional data, but only capture one view-angle of the entire earth. With different satellites from the American weather service, the European EUMETSAT and Japanese Himawari satellites, the entire earth is covered. Geostationary satellites are 40.000 kilo-meter away from earth to maintain a fixed position and observe earth with with ~1km data resolution every 15 minutes.
Satellites however do not provide data for different atmospheric levels. (Technically modern satellites now use radar observations for wind speeds at different altitude levels, but they need to be closer to earth and have to orbit earth multiple times a day.)
Airplane data can provide accurate measurements from different altitudes as well as sounding measurements from weather balloons. Especially in the Atlantic Ocean, this could improve data assimilation.
In reality, the German Weather service DWD just stopped using airplane measurements, because measured data shows large discrepancies between airplane data and soundings from weather balloons.
A newly developed diagnostic system in the data assimilation section showed that there is a large discrepancy between temperature measurements from the AIREP system and corresponding radiosonde temperature observations. A reason for that is a large temperature bias of the AIREP temperature observations, which cannot be corrected by the aircraft bias correction system due to a missing entry in the Bufr decoding.
Therefore, the temperature measurements of the AIREP system will be set passive, starting from Wednesday 13 October 2021 09 UTC. This was successfully tested in a month long experiment showing a slightly positive impact for Europe after 48 hours.
Source: DWD Operational NWP System Newsletter. No online version available.
Without airplane data, weather-forecasts for Europe actually improve!
Hopefully data discrepancies from airplane data are soon resolved and can provide important data for weather-model assimilation. I am curious how weather forecasts quality changes, once airplane data is reintroduced.
An important takeaway is to always validate data and ensure quality over time! Especially for measurements I often observe larger errors because of weather station placement or hardware sensor faults.
DWD weather models are an important part of Open-Meteo APIs and provide high resolution data for Central in Europe. We highly appreciate the level of professionalism of the German Weather Service and constant quality improvements of their weather forecasts.
I am looking forward to report the next developments in the field of numerical weather predictions!