The seminar is recorded and can be found here and includes the presentation.
Abstract
Meteorological reports from aircraft are an important source of data for Numerical Weather Prediction (NWP). From a 'no aircraft' experiment run in 2019 the largest impact is in the upper troposphere at short-range. The number of aircraft reports reduced by about 75% between mid-March and mid-April 2020 due to the disruption caused by Covid-19, recovering to almost 50% of normal by July 2020. Despite the hiatus it is difficult to see any evidence of a drop in forecast skill - partly because forecast skill is very variable, showing both day-to-day noise and lower frequency trends. Satellite data are very important for NWP and several new sources became available during 2020.
Among the measures taken at ECMWF to mitigate the data loss was the assimilation of European Mode-S aircraft winds - these are still very dense and only 5% of them are used. The use of Mode-S winds improves the 12-hour forecasts over Europe as measured against radiosondes - with the largest impact around 250 hPa again. A separate change to try to correct wind direction errors in a subset of B787 reports was also introduced. Although only about 10% of the affected winds are in the southern hemisphere the forecast improvement is largest there.
Aircraft temperatures are biased (usually between 0.3 and 1 degree too high) and NWP centres have to bias correct them. More metadata, such as aircraft type, would help with this and other quality issues. Finally, efforts to "fill in the gaps" - provide aircraft reports in currently data sparse regions were mentioned.