AMDAR (Aircraft Meteorological DAta Relay) automated weather reports from commercial aircraft provide an increasing amount of input data for numerical weather prediction models. Previous studies have investigated the quality of AMDAR data. Few of these studies, however, have revealed indications of systematic errors dependent upon the aircraft type. Since different airlines use different algorithms to generate AMDAR reports, it has remained unclear whether a dependency on the aircraft type is caused by physical properties of the aircraft or by different data processing algorithms. In the present study, a special AMDAR dataset was used to investigate the physical type-dependent errors of AMDAR reports. This dataset consists of AMDAR measurements by Lufthansa aircraft performing over 300 landings overall at Frankfurt Rhein/Main (EDDF/FRA) on 22 days in 2004. All of this data has been processed by the same software, implying that influences from different processing algorithms should not be expected. From the comparison of single descents to hourly averaged vertical profiles, it is shown that temperature measurements by different aircraft types can have systematic differences of up to 1 K. In contrast, random temperature errors of most types are estimated to be less than 0.3 K. It is demonstrated that systematic deviations in AMDAR wind measurements can be regarded as an error vector, which is fixed to the aircraft reference system. The largest systematic deviations in wind measurements from different aircraft types (more than 0.5 m s −1 ) were found to exist in the longitudinal direction (i.e. parallel to the flight direction).

This is one of a series of papers on the Askervein Hill Project. It presents results from the Askervein 1982 and 1983 experiments in the form of vertical profiles of mean wind and turbulence integral statistics at upwind reference locations and at two hilltop sites. The data were obtained from a variety of sensors including sonic, Gill UVW and cup anemometers mounted on 50,30,17 and 10 m towers and TALA kite systems. Comparisons with numerical-model predictions are discussed.

Two state-of-the-art, high-resolution, in situ turbulence measurement systems, which can be deployed at altitudes well above the atmospheric surface layer, are compared: the Tethered Lifting System (TLS) of the Cooperative Institute for Research in Environmental Sciences (CIRES) at the

This is one of a series of papers on the Askervein Hill Project. It presents results on the variations in mean wind speed at fixed heights (Az) above the ground from linear arrays of anemometer posts and towers. Most of the data are for AZ = 10 m but some are for AZ = 3 m. Selected and directionally grouped data from the 55 'Mean Flow' runs are presented together with mean flow data from Askervein '83 'Turbulence' runs. Comparisons are made between the data and guideline estimates of fractional speed-up ratio at hilltop locations and between the data and MS3DJH/3 model predictions along the tower lines. There is good agreement in most cases.

At Frankfurt airport (EDDF) vertical soundings of the lower atmosphere from two independent sources are available. One of them is a wind and temperature profiler (wind temperature radar and radio acoustic sounding system, WTR/RASS) located at the western end of the main pair of runways. The second source is aircraft meteorological data relay (AMDAR), i.e. measurements operationally collected by approaching and departing aircraft. Together, both offer a rare opportunity to compare the performance of these widely used systems. We use 1 year of continuous data to calculate systematic temperature and wind vector differences between both measurement systems. The differences show a clear seasondependent structure in conjunction with typical inversion heights. Possible causes for this behaviour are discussed. Second, we compare the ability of both systems to identify inversion and wind-shear layers above the airport. AMDAR-detected layers are typically higher than wind profiler detections. The layer base is usually detected with more agreement than the top. The mutual probability of detection of inversions is found to be mostly between 40 and 60%.