Finding Boundary Layer Top: Application of a Wavelet Covariance Transform to Lidar Backscatter Profiles

Abstract

Several recent studies have utilized a Haar wavelet covariance transform to provide automated detection of the boundary layer top from lidar backscatter profiles by locating the maximum in the covariance profiles. This approach is effective where the vertical gradient in the backscatter is small within and above the boundary layer, and where the inversion is sharp and well defined. These near-ideal conditions are often not met, particularly under stable stratification where the inversion may be deep and is sometimes ill defined, and vertical gradients are common. Here the effects of vertical gradients and inversion depth on the covariance transform are examined. It is found that a significant dilation-dependent bias in the determination of the boundary layer top may result when using the published method. An alternative approach is developed utilizing multiple wavelet dilations, and is capable of identifying both the upper and lower limits of the backscatter transition zone associated with the inversion while remaining insensitive to mean vertical gradients in the background signal. This approach enables more detailed information on the small-scale structure of the inversion and entrainment zone to be retrieved than is possible using existing techniques.

Black carbon solar absorption suppresses turbulence in the atmospheric boundary layer

“…The atmospheric boundary layer height was determined using the MiniMPL micro pulse lidar (Sigma Space) (18). To distinguish aerosol features confined to the surface mixed layer from elevated aerosol plumes, an upper bound was imposed on the automated algorithm based on subjective analysis of the backscatter profiles ranging from 900 m to 1,700 m, depending on the day.…”

Automatic detection of atmospheric boundary layer height using ceilometer backscatter data assisted by a boundary layer model

“…The wavelet analysis, being based on an integral rather than a differential method, is more robust and less sensitive to noise, also as a result of its multi-scale approach. Conversely, the wavelet transform suffers of a large blind zone, which is half of its maximum extension, as already observed in Brooks (2003). This makes it impossible to detect low-altitude inflection points, which are instead very common at night-time, as can be seen in Figure 1, for example.…”

“…The S × R 2 inflection points, which are tracers of atmospheric stratifications, are searched employing both the derivative of the signal calculated by finite differentiation (gradient method) and the WCT, which consists of the convolution of the signal and a wavelet function. The implemented WCT algorithm employs a step-like Haar wavelet, and follows the scheme previously described in the literature (Brooks, 2003;Haij et al, 2007;Morille et al, 2007). This wavelet is characterized by the size of its non-null part (usually referred to as dilation), which determines the scale of the features that can be revealed by the WCT.…”

“…Technically, the processing is performed either by direct numerical calculation, finding the relative minima of the first derivative of the range-corrected signal, or by means of the discrete wavelet transform (DWT), also sometimes referred to as wavelet covariance transform (WCT) method. The latter, initially employed by Ehret et al (1993), Wang and Lenschow (1995), Kiemle et al (1995), Davis et al(1997), Russell et al (1998), Sullivan et al (1998), Cohn et al (1998), Cohn and Angevine (2000), Davis et al(2000), Brooks (2003), and more recently by Morille et al(2007) and Martucci et al (2010aMartucci et al ( , 2010b, consists in the convolution of the vertical signal with a step function, which is able to detect local discontinuities in the backscatter profile. For each column these signal-processing techniques identify multiple points along the vertical.…”

Overview of the Earthcare L2 Lidar Retrieval Chain

“…The ATLID standalone layer products A-CTH and A-ALD are prerequisites for the corresponding synergistic layer products. The A-CTH algorithm makes use of a combined wavelet covariance transform (WCT) and threshold approach (e.g., [6], [7]) to determine the uppermost cloud top height along the ATLID track with a resolution of 1 and 10+ km. The latter, coarser resolution is needed to identify optically thin clouds, e.g., subvisible cirrus.…”