Key Analysis Errors and Airborne Wind Lidar Observations

Kurzfassung

Uncertainties in the initial conditions are a major source of errors in numerical weather forecasts. An iterative procedure that uses the adjoint version of the forecast model to minimize the short-term forecast error yields so-called key analysis errors. Under the assumption that the growth of forecast errors is mainly due to errors in the forecast initial state rather than model errors, the key analysis errors are expected to represent substantial parts of the analysis error that is responsible for a poor forecast. Furthermore, the identifcation of the analysis error can be achieved by monitoring the differences of observations and analysis fields (analysis departures). Airborne Doppler wind lidar observations over the Northern Atlantic THORPEX Regional Campaign (A-TReC) are analysed to test these considerations. The wind observations, taken with the DLR (Deutsches Zentrum für Luft- und Raumfahrt, Oberpfaffenhofen) 2-µm Doppler wind lidar, were passively monitored and actively assimilated in experiments using the ECMWF global model to form the basis for the computation of analysis departures and analysis differences. The aim of this thesis is to gain a further understanding of the structure of key analysis errors and to investigate the question how well key analysis errors are related to analysis errors. The results confrm that analysis departures and key analysis errors optimized for both the Northern hemisphere and a targeted area are not correlated. Quantitative and qualitative comparisons show substantial differences of key analysis errors and analysis departures. It is also shown that similar discrepancies are found between key analysis errors and analysis differences.