|Title||Atmospheric motion vectors from model simulations. Part I: Methods and characterisation as single-level estimates of wind|
|Publication Type||Technical memorandum|
|Secondary Title||ECMWF Technical Memoranda|
|Authors||Hernandez-Carrascal, A, Bormann, N, Borde, R, Lutz, H-J, Otkin, J, Wanzong, S|
The main objective of this study is to improve the characterization of Atmospheric Motion Vectors (AMVs) and their errors to improve the use of AMVs in Numerical Weather Prediction (NWP). AMVs are estimates of atmospheric wind derived by tracking apparent motion across sequences of satellite images, and they tend to exhibit considerable systematic and random errors and geographically varying quality. These errors can arise in the AMV derivation or the interpretation of AMVs as single-level point observations of wind. An important difficulty in the study of AMV errors is the scarcity of collocated observations of clouds and wind. To overcome that difficulty, this study uses a simulation framework: geostationary imagery for Meteosat-8 is generated from a high resolution NWP model simulation (performed with the WRF regional model with a nominal horizontal resolution of 3 km), and AMVs are derived from sequences of these simulated images. The NWP model provides the "truth" with a sophisticated description of the atmosphere. The study considers infrared and water vapour AMVs from cloudy scenes. This is the first part of a two-part paper, introducing the simulation framework and providing a first evaluation of the simulation. The key results are: 1) cloud structures in the WRF simulated imagery are generally realistic, although there are some unrealistic aspects during the first part of the study period, likely related to the spin-up, 2) the statistics comparing the simulated AMVs with the true model wind show characteristics that are broadly similar to statistics comparing real AMVs and NWP short-term forecasts, although differences appear to be larger, and 3) there is evidence for significant spatial, temporal and vertical error correlations, with the scales for the spatial error correlations broadly matching similar estimates for real data. The second part focusses on observation operator aspects.