Enabling Assimilation of GNSS RO Data in the Lower Troposphere and Boundary Layer

The National Academy of Sciences (NAS) and NASA have identified profiling the planetary boundary layer as a key goal necessary for improving our understanding and forecasting capabilities for weather, climate, and air quality. GNSS radio occultation (RO) has a unique set of features well suited to profiling the shallow boundary layer from orbit which suggest RO will play an important role in achieving these goals. These features include very high vertical resolution, high precision, all weather profiling, insensitivity to the underlying surface, and unbiased global coverage.

While RO has these features, numerical weather prediction (NWP) systems are making limited use of RO profiling in the lower troposphere and boundary layer (LTBL) over much of the globe. A key reason is assimilation of RO bending angles in the LTBL becomes numerically unstable when the vertical gradients in refractivity become very large such as across the sharp transitions in moisture between the boundary layer and overlying free troposphere. As such, NWP systems avoid assimilating RO in LTBL to avoid it crashing their data assimilation system

With funding from the US Air Force, we are implementing a solution in JEDI to this present NWP limitation. The solution utilizes very high SNR GNSS RO observations and a combination of several key RO related capabilities developed over past 20+ years.

This effort is focused in particular on a solution that works in ducting conditions, where an ill posed problem exists where the one observed bending angle profile is consistent with a continuum of refractivity profiles. The solution to this ill posed problem combines the method of Xie et al. (2006) to recover the continuum of potential refractivity profile solutions and then uses grazing reflected bending angles as identified by Aparicio et al (2018) and demonstrated by Wang et al. (2022) as the additional constraint to identify and select the optimum refractivity profile from the continuum. High SNR RO observations are critical here in order to routinely measure the grazing reflected signals.

To avoid the instability associated with assimilating bending angles, RO assimilation will be accomplished via the nonlocal pseudo excess phase forward operator (FO) of Sokolovskiy et al (2005) which is based on the nonlocal refractivity FO of Syndergaard et al (2003, 2005). We will implement the nonlocal pseudo excess phase FO in JEDI to enable assimilation of the RO data. We will derive nonlocal pseudo excess phase estimates from the observations that will be assimilated.

A key aspect of this implementation is using high SNR RO observations to acquire unbiased and low noise bending angle profiles in the LTBL complete with accurate estimates of the bending angle uncertainty together with very high vertical resolution profiles particularly for identifying the height of the boundary layer.

We will describe our approach and present a progress report on this development.