In principle, MLH detection is a pattern recognition problem. The basic assumption which is usually made is that the vertical distribution of aerosol can be used as a tracer for finding boundaries. The absolute value of the backscatter is typically not needed since the relevant information seems to be completely coded in the gradient (but possibly of different orders) of the backscatter profile. Currently, two major types of algorithms for MLH detection from lidar or ceilometer exist: (a) The peak technique, which was invented for the detection of the top of multiple aerosol layers from backscatter profiles and which is based on the analysis of collocated minima of the backscatter gradient and maxima of the backscatter variance. (b) The wavelet method, which utilizes the wavelet transform of the backscatter profile. This method has gained a great amount of popularity during the last years. Typically, the Haar wavelet transform is used because it is easy to implement and a powerful gradient locator and therefore a very promising mathematical tool; but recently published wavelet-based methods do not take full advantage of the wavelet theory. Here, we present first results of our efforts to develop an advanced version of the wavelet algorithm and, thus, a reliable MLH detector.
First, the new MLH retrieval technique is described, demonstrated in a case study, and compared to results obtained with the standard peak technique. Then, a comparison to MLH derived from radiosonde data will be presented. Emphasis is also put on the assessment of the new ceilometer in comparison with other remote-sensing systems like a Ka-band cloud radar. Furthermore, a first impression of the spatial MLH variability over Germany is given by showing results from different ceilometer network sites. Finally, the potential for further improvements of the new profiling technique using the new type of ceilometers is discussed.