Validation at high frequency of wind field reconstruction from scanning lidar using Gaussian processes
report
During 2017, a novel method was developed – based on a machine learning approach called Gaussian Process Regression – for using raw Lidar data to reconstruct instantaneous 3D wind fields. This method was successfully validated using a Windcube v2 ground-mounted static beam Lidar, against 10-minute averaged data from several cup anemometers at different heights on a met mast located on a site with “simple” terrain.
The current report summarises an innovation project funded by ECN part of TNO to collect a new data set using its Windcube 200s scanning Lidar, in order to: 1 validate against a 3D sonic anemometer at 1Hz frequency for clean and complex wind conditions; 2 understand how the method performs when applied to a different Lidar design and scanning pattern. Apart from implementing a method to filter laser pulse reflections from the met mast, no changes were required to the method presented in previous reports.
The current report summarises an innovation project funded by ECN part of TNO to collect a new data set using its Windcube 200s scanning Lidar, in order to: 1 validate against a 3D sonic anemometer at 1Hz frequency for clean and complex wind conditions; 2 understand how the method performs when applied to a different Lidar design and scanning pattern. Apart from implementing a method to filter laser pulse reflections from the met mast, no changes were required to the method presented in previous reports.
TNO Identifier
865673
Publisher
TNO
Collation
19 p.
Place of publication
Petten