Application of Gaussian Processes to Dual-Doppler LiDAR scanning measurements for high frequency wind field reconstruction
report
The work here presented aims to demonstrate the potential of applying Gaussian Processes regression, a machine learning technique, for the purposes of extracting more information from scanning LiDARs measurements. With such goal in mind, two major tasks constitute the bulk of the work here presented: (1) application of Gaussian Processes to one single scanning LiDAR for data interpolation and (2) application of Gaussian Proccesses to two scanning LiDARs operating in dual-doppler mode for wind field reconstruction. For such tasks, data measurements from a test campaign at Bremerhaven test field from Fraunhofer IWES are used. These were collected from the 14th of March until the 9th April. The data measurements collected and used for this work include data from 2 scanning LiDARs operating in dual-doppler mode, a cup anemometer positioned at 55 meters of height and a wind vane positioned at 110 meters of height, the last two installed in a IEC-compliant met mast.
TNO Identifier
960965
Publisher
TNO
Collation
49 p.
Place of publication
Petten, The Netherlands