15th International Winds Workshop April 12 - 16, 2021 Virtual (13 UTC Start Time)

An important diurnal cycle in convection, cloudiness and surface temperature exists for all regions of the tropics. It is currently not fully clear how these tropical specificities affect AMVs extracted from satellite imagery. Recent reports (e.g. Horvath et al, 2017; Warrick, 2016) revealed the existence of a positive Observation-minus-Background (O-B) speed bias in the tropical region of the upper troposphere for most satellite-channel combinations from both geostationary and polar-orbiting satellites. O-B speed biases are commonly explained by erroneous altitudes assigned to AMVs. However, fast speed biases are found for AMVs that are set already very high in the troposphere and it does not appear very realistic to consider that they are set too low.

A study is presented that aims exploring how the dynamics and physics underlying the tropical atmosphere affect AMVs extracted from satellite imagery. Monthly statistics of 1 yr of data of AMVs derived by EUMETSAT from IR imagery from Meteosat-10 (Met10) and Metop-A/B satellites against hourly ECMWF forecast wind fields are derived. Furthermore, the Meteosat-10 AMV performance over desertic areas is studied in more detail. Results reveal a good agreement between satellite and model winds for pressures ≤ 250 hPa. Between 300-500 hPa, altitudes assigned too low lead, in conjunction with vertical wind shear and generally fast winds, to AMVs being frequently more than 5 ms^-1 faster than ECMWF. In contrast, height assignment errors unlikely explain why Dual-Metop AMVs are faster than model winds over the Boiler-Box region (Indonesia, Papua-New Guinea). AVHRR brightness temperature fields indicate that due to the large temporal gap of 50 minutes between two images and due to the strong convection that alters the shape of clouds, the feature to be tracked is difficult to relate accurately between two AVHRR images. The brightness temperature fields of AVHRR image pairs further reveals generally low spatial correlation between target box and its location in the search area. Above findings are further investigated using wind profiles from HLOS ADM-Aeolus. A brief discussion on recommendations on AMV extraction in tropics based on such cases will be given.