Presentations and Publications

Stanier, C., Schoenfelder, J., Yarker (Brown), M.  Evaluation of the Vaisala CL31 ceilometer as a tool for boundary layer characterization within carbon cycle studies.  Report to the NOAA Global Monitoring Division and Vaisala.  April 2009.

During the period June 18 to July 18, 2008, a Vaisala CL31 ceilometer was operated at the Southern Great Plains (SGP) ARM site in Oklahoma. Then the same instrument was shipped to Iowa and operated at the WBI (West Branch Iowa) tall tower CO2 characterization site. The operation period at WBI was from July 25, 2008 to Aug 26, 2008. The backscatter product from the CL31 was compared qualitatively to three operational measurements at SGP – the previous generation Vaisala CT25K ceilometer, the Micropulse LIDAR, and the four per day rawinsondes. Boundary layer heights were determined manually from inflection points in potential temperature and relative humidity in the rawinsonde data. These were compared to software-determined mixed layer heights from the CL31 using two approaches: Vaisala’s MLH software package version 3.0 (a MATLAB executable with extensive graphical user interphase)
and a simple boundary layer detection program that finds the point where relative backscatter decays quickest as a function of height.

Qualitative agreement between the three remote sounding instruments was excellent. The spatial and temporal resolution of the CL31 was superior to that of the CT25K, and the signal to noise ratio appears to stay stronger to higher altitudes with the CL31 (compared to the CT25K). At altitudes less than 100 m, the CT25K and CL31 are qualitatively different, with the CT25K usually having a backscatter peak highest at around 80 m, and the CL31 product smoothly increasing up to the a maximum at the surface.

Both algorithms performed well for matching rawinsonde-derived PBL features between 0.5 and 2 km. Ceilometer-derived boundary layer heights were within 15% of the sonde-derived heights in about 60% of the examined cases. For matching sonde features less than 0.5 km (these were not typically mixed layer heights, but rather the height of stable atmospheric boundary layers) only the Vaisala algorithm was used. In this height range, the Vaisala algorithm tended to cluster the height around a smaller range (120-220 m) than determined by the sondes (40-350 m). Therefore, relative error in the boundary layer height is high. Although there was high correlation between rawinsonde-derived PBL features and ceilometer-based features, the multilayer structure of the atmosphere (corresponding to 2-4 different layers in any given launch) will make interpretation of CL31 data alone difficult.

At West Branch, Iowa, the comparison was made between the CL31 product time series (backscatter curtain plots and mixed layer height) and the CO2 concentrations measured at the three tower elevations of 31, 99 and 379 meters above ground. The time when the mixed layer height grows to these tower heights is clearly observed in the CO2 record during the growing season. The time of growth to 379 m is tightly correlated with the CL31 mixed layer height. The time of growth to 99 m is not well correlated with the CL31 mixed layer height, although modification of the processing algorithm for the CL31 could probably be done to improve this agreement.

With respect to carbon cycle studies, there does seem to be a potential to constrain both nocturnal and daytime boundary layer heights at individual sites using ceilometers. For this to be done operationally, substantial work would be needed (possibly on a site-by-site basis) to fine tune the software so that the appropriate boundary layer features of interest could be identified, especially during periods with multiple layers.