Treffer: Hydraulic Data Analysis Using Python

Acoustic Doppler velocimeter (ADV) data is prone to high uncertainty in measurement. In this thesis, technical literature that proposes data analysis methods to reduce error effects is reviewed, and subsequently, three methods are implemented using the programming language Python. The reduction of uncertainty in measurement is evaluated by categorising statistical parameters and analysing time-series and Kolmogorov energy spectra for 160 ADV samples in turbulent flow. The results show that out of the examined data analysis methods, kernel density estimation despiking in combination with lowpass Butterworth filtering is the most promising way to reduce the uncertainty in measurement. Furthermore, a procedure to find the optimal sampling time for ADV measurements is realised. The implementation shows that statistical equivalence testing is adequate for finding the optimum sampling time. Still, the procedure needs further development to provide significance regarding higher statistical moments. Ultimately, a systematic workflow for handling ADV data is proposed.