Refinements of the algorithms needed to retrieve atmospheric vertical thermodynamic profiles from GNSS radio occultations

In this work we consider some refinements to the Boundary Profile eValuation (BPV) method, developed by Vespe et al. (2004), Vespe and Persia (2006), Vespe (2016), Andrisani and Vespe (2020) in order to retrieve atmospheric thermodynamic profiles from Radio Occultation (RO) observations. Former BPV models solved typical rank deficiencies in treating RO data by applying parametric dry refractivity models for the atmosphere, such as the CIRA86aQ or Hopfield (1969). The involved parameters were selected by fitting those dry models with the refractivity observed in the stratosphere, where water vapor concentrations are negligible, with the usual Least Square Error (LSE) method. Dry refractivity is then extrapolated down to the Earth surface. Nevertheless, unphysical occurrence of negative values of humidity concentrations, typical for the so called "simple models" (Kursinski and Kursinski, 2013) can occur in the lower troposphere. This problem is solved with a non-negative residuals constrained fitting for the total refractivity, obtained with different mathematical methodologies. Other refinements include the departure from the ideal state equation for a more realistic state equation concerning compressibility factors for the dry and wet components of the atmosphere. An application to 848 RO observations from COSMIC_1 space mission, covering tropical and middle latitudes during the months of January, March and May 2009, together with a statistical comparison with other atmospheric models and experimental data, was made in order to estimate the accuracy of the BPV method. The results show a description about the thermodynamic state of the atmosphere in line with other recognized atmospheric models, in particular for what concerns the tropics and the southern middle latitudes. (C) 2020 COSPAR. Published by Elsevier Ltd. All rights reserved.