Task 116

Automating Boundary Layer Detection for Aerosol Lidar

Principal Investigator(s):

Z. Li


V. Sawyer


J. Richards/J. Welton

Last Updated:

October 26, 2012 15:25:27

Description of Problem

The planetary boundary layer (PBL) varies in depth on a time scale of minutes to hours, with implications for aerosol transport, surface air quality, and radiative forcing. Unfortunately the most direct method for observing the PBL top height, via thermodynamic profiles from radiosonde launches, is available only four times per day even during intensive campaigns and less frequently than that during normal operational use. The timing of the launches seldom corresponds to the extremes of the diurnal cycle. Because a higher temporal resolution for PBL measurements would be valuable to modeling efforts, remote sensing methods are an important potential source. In order to determine whether aerosol profiles detected by micropulse lidar can serve as effective proxies of the thermodynamic structure, PBL heights derived from them must be compared to PBL heights taken from radiosonde and other thermodynamic profiles, such as AERI retrievals. All three sets of measurements are available at the ARM SGP site; intensive collocated radiosonde and MPL data come from the ICEALOT research cruise of March-April 2008.

Scientific Objectives and Approach

A wavelet covariance transform technique (Davis et al. 2000; Brooks 2003) is combined with a curve-fitting iterative process (Steyn et al. 1999, Hägeli et al. 2000) to detect the PBL top height in backscatter profiles from ground-based aerosol lidar such as those at ARM sites and in the MPLNET group. The resulting PBL heights were compared to PBL heights taken from Liu and Liang (2010), who analyzed the radiosonde record at the SGP site over several years’ worth of data. In addition, the algorithm has recently been adapted to work with AERI retrievals of virtual potential temperature. Some changes to the algorithm are needed to reduce errors and artifacts in the resulting PBL time series. The accuracy of the results will vary by time of day and the stability profile of the lower troposphere.


The combined algorithm is more sensitive to small-scale changes in the PBL height over time, while less susceptible to errors caused by the extraneous backscatter signals of high clouds, elevated aerosol plumes, etc. than the wavelet covariance technique alone. It remains suitable for automated PBL detection, requiring little prior knowledge or computational resources. It can be shown that while Steyn et al. (1999) used the iterative method to find entrainment zone depth as well as PBL height, the drop in aerosol concentration at the boundary does not necessarily correspond to the depth of the temperature inversion found in thermodynamic profiles. A journal article is in the works.

Other Publications and Conferences

Sawyer, V.R., Z. Li, and E.J. Welton, 2011. Aerosol lidar observations of the planetary boundary layer and warm conveyor belt. Gordon Research Conference, Waterville, ME.

Sawyer, V.R. and Z. Li, 2011. Boundary layer and entrainment zone observations using MPLNET lidar. ASR Science Team Meeting, Annapolis, MD.

Sawyer, V.R., Z. Li, and E.J. Welton, 2011. Validation of boundary layer detection by ground-based aerosol lidar. IYC Ozone Meeting, Washington, D.C.

Sawyer, V.R., Z. Li, and E.J. Welton, 2012. Detection of boundary layer and entrainment using ground-based aerosol lidar. AMS Annual Meeting, New Orleans, LA.

Sawyer, V.R., Z. Li, and E.J. Welton, 2012. Comparison of boundary layer detection between instruments at the Southern Great Plains site. ASR Science Team Meeting, Arlington, VA.

Task Figures

Fig. 1 – Comparison between combined-algorithm lidar PBLs and those from Liu and Liang (2010), during the period 2003-2004 at the ARM SGP site. Cases are classified by the stability regime of the lower troposphere: stable PBLs often occur below the minimum range detectable by lidar, while neutral and convective conditions allow for more accurate results.