Presentations

Abstract

The existing IEC-standard methods for calibrating and classifying lidars are too time-consuming and systematically overestimate lidar uncertainties. These shortcomings are largely due to the calibration reference instruments: anemometers mounted on meteorological masts. The input signals in field calibrations are real wind conditions, including the uncertainties and influences of the meteorological mast and surrounding terrain. Indeed, the largest contributor to the final, derived lidar uncertainties is the inherited uncertainty from the reference instruments and its installation. For comparison, the reference uncertainty in an anemometer wind tunnel calibration is typically 0.05 m/s, or 0.7% for a 7.5 m/s wind and the calibration process takes a few hours. For a lidar, a typical reference uncertainty (from the mast) is 1.7%, and the calibration process takes weeks to complete due to the natural variation of the wind. In this presentation, we demonstrate a new methodology for lidar calibration that uses a fiber optic bench as a synthetic wind tunnel for lidars called Simulation of the Atmosphere with Fiber Optics (SAFO). Using the SAFO bench, the lidar line of sight (LOS) signal is shifted using optical components and backscattered from a 500 m fiber spool, effectively mimicking the backscatter from a uniform, low turbulence wind field. The lidar calibration process takes two days and covers all possible radial velocities and lidar signal conditions for each LOS. The electric and optical signals are SI traceable to international standards and have reference uncertainties of 0.02 m/s or 0.3% for a 7.5 m/s wind, much smaller than the uncertainties inherited from reference anemometers. The horizontal wind speed uncertainties are derived via Monte Carlo uncertainty propagation using the results of the SAFO bench calibration, and other lidar optical properties as inputs. This builds on uncertainty propagation methods in the IEC 61400-50-3 and the framework of IEC 61400-50-1 Section 7.3, where a sensor response is input to a physical simulation to derive a comprehensive device classification. We demonstrate that the SAFO and Monte Carlo framework can replicate known lidar sensitivities compared to a virtual reference, including (1) wind field reconstruction, (2) volume averaging effects, and (3) low availability conditions. We compare the Monte Carlo results to IEC 61400-50-2 Classification sensitivities and Calibration uncertainties for six systems. The new method yields uncertainties roughly 50% lower than the IEC field calibrations and is 10 times faster. This new technology has the potential to streamline lidar deployments and reduce the sensor uncertainty contribution to wind resource assessment, while reducing costs. This calibration and classification process is more rigorous than today’s lidar calibration methods as identical conditions can be applied to every lidar and can cover a wider range of atmospheric conditions.