Effects of Spatial Averaging in Vertical Velocity Statistics from the ARM Doppler Lidars

Description

The U.S. DOE ARM Facility currently operates nine coherent Doppler lidars (CDLs) at various sites, including five systems at the Southern Great Plains site. These systems are configured to spend the bulk of their time acquiring high-temporal-resolution (~1 Hz) profiles of clear-air vertical velocity with a height resolution of ~30m. In addition to these raw measurements, 30-minute-averaged profiles of vertical velocity variance, skewness, and kurtosis are also computed and made available to the science community through the ARM web site. The interpretation of these lidar-derived statistics is complicated somewhat by the fact that the raw measurements represent spatially averaged quantities due to the finite dimensions of the lidar pulse and the range gate width. This spatial averaging results in variance estimates that are biased low when compared to variances computed from point measurements such as sonic anemometers. In this study, we compare vertical velocity variances from one of the ARM CDLs to variances from nearly collocated tower-mounted sonic anemometers in order to quantify the bias in the lidar-derived estimates. This comparison is performed using data that were acquired during the eXperimental Planetary boundary-layer Instrument Assessment (XPIA) field campaign at the Boulder Atmospheric Observatory (BAO) in March and April of 2015. During XPIA the ARM Doppler lidar that was formerly operated at the Tropical Western Pacific site in Darwin, Australia was deployed next to the 300-m tower at the BAO site, and the tower was instrumented with sonic anemometers at six levels. This experimental set-up enabled the estimation of the CDL’s spectral transfer function from the ratio of the CDL to the sonic turbulent power spectral densities. The transfer function is then used to correct for the effects of spatial averaging in the CDL-derived variance estimates. Mean variance profiles from the sonic anemometers are compared to the corrected and uncorrected mean variance profiles from the CDL.