Radio Occultation Observation Operators for Data Assimilation

In the current numerical weather prediction data assimilation systems, the radio occultation measurements are available in the form of bending angles or refractivity. The forward observation operators are in most cases one dimensional and implemented under the assumption of spherical symmetry neglecting the horizontal gradients of the atmospheric refractivity. There are, however, operational implementations of 2D and 3D operators. In this study, we propose to implement and evaluate 3D and 1D observation operators for the Spire numerical weather prediction model, that consider horizontal characteristics of the atmosphere and include the ray-tracing model. The novelty of this approach regarding the 1D operator consists in using the ray-tracing model. Current versions of 1D operator use the discretized form of the Abel integral, which may result in inaccuracies in the presence of atmospheric wave-guides. This approach makes the 3D and 1D operators consistent with each other. Using radio occultation observations from the Spire satellite constellation, we will calculate specific statistics between real and modeled observations computed by the 1D and 3D forward operators to search for regimes where accuracy improvements are notable. The final goal is to derive a computationally efficient and accurate hybrid (1D+3D) radio occultation operator. Benefits of employing the 3D operator are expected for occultation points situated at elevations lower than 10km. At higher altitudes, the 1D and 3D operators should produce modeled observations with similar accuracy.