Removing the a Priori Effect from a Lidar Optimal Estimation Method Water Vapor and Temperature Retrieval

The Optimal Estimation Method (OEM) has a long history of success for retrievals of atmospheric composition from passive remote sensing instruments. Recently, Povey et al. (2014) and Sica and Haefele (2015, 2016) also implemented the OEM retrievals for lidar measurements of aerosol, temperature, and water vapour mixing ratio.

The OEM retrieval is uses the raw (uncorrected) lidar measurements and background knowledge, called the a priori. The dependence of the OEM retrieved profile on the original measurements depends on the signal-to-noise ratio (SNR). When the SNR is high, the retrieval is entirely dependent on the measurements and there is no contribution from the a priori. The contribution of the a priori increases when the SNR drops and there is increasing uncertainty in the retrieved parameters. The SNR is particularly low when the background is high, such as during the daytime for water vapor retrievals.

We have developed a method to remove the effect of the a priori from the RALMO water vapor and PCL temperature retrievals using an information-centered approach. This method uses the Von Clarmann (2009) approach to compute a “coarse” retrieval grid on which the retrieval is recalculated, removing the a priori contribution from the retrieval. Measurements from two lidars were used in this study, the RAman Lidar for Meteorological Observation (RALMO) at the MeteoSwiss Research Station in Payerne, Switzerland and the Purple Crow Lidar (PCL) located in London, Canada. The PCL Rayleigh signals and RALMO water vapor signals were used to retrieve the OEM temperature and water vapor profiles following the same procedures as in Sica and Haefele (2015,2016) on the original “fine” grid. We find this procedure increases the retrieval altitude by 1 km during the daytime for water vapour retrievals, from about 4 to above 5 km altitude. Similarly, this method improves the validity of the measurements for the PCL temperature in the regions in the altitudes where the SNR is low by 3 km.