Abstract/Summary

This thesis describes the development, characterisation and use of a dataset of measurements made using 51 radiosondes equipped with accelerometers to measure atmospheric turbulence. Atmospheric turbulence, especially Clear-Air Turbulence (CAT) is hazardous to aircraft as it cannot be observed in advance. Pilots and passengers rely on CAT forecasts, which at best, are correct 60-70% of the time. The reason for this moderate performance in turbulence forecasts is due to a lack of quantitative unbiased observations needed to improve the turbulence theory. This work seeks to improve understanding of turbulence through a standardised method of turbulence observations that span the entire troposphere. To achieve this a sensing package is developed to measure the acceleration of the radiosonde as it swings due to its carrier balloon being agitated by turbulence. The accelerometer radiosonde is then compared against multiple turbulence remote sensing methods to characterise its measurements. From a comparison with a Doppler lidar in the boundary layer a relationship in terms of the eddy dissipation rate, a meteorological measure of turbulence, is found. A further relationship is found when compared with the spectral width of an Mesospheric Stratospheric and Tropospheric (MST) radar. The full dataset of accelerometer sonde ascents is analysed and with information from instrumental comparisons a standard deviation of 5 m s−2 is defined as a threshold for significant turbulence. The dataset spans turbulence generated in meteorological phenomena such as jet streams, clouds and in the presence of convection. The analysis revealed that 77% of observed turbulence could be explained by the aforementioned phenomena. In jet streams turbulence generation was often caused by horizontal processes such as deformation. In the presence of convection turbulence is found to form when CAPE > 150 J kg−1. Deeper clouds were found to be more turbulent due to the increased intensity of in-cloud processes. The accelerometer data were used to verify the skill of turbulence diagnostics, in order to assess which diagnostics are best at forecasting turbulence. It was found that turbulence diagnostics featuring the wind speed, deformation and relative vorticity advection predicted turbulence best. This work provides a new, safe and inexpensive method to retrieve in-situ information about the turbulent structure of the atmosphere. It can inform the aviation industry on where turbulence is generated and assess which are the most skilful diagnostics to predict this.