Impact of Two Radio Occultation Constellations on Numerical Weather Prediction: Polar‐Orbits Only and Mixed Polar‐Equatorial Orbits

This study explores the relative impact of two orbital configurations for Global Navigation Satellite System radio occultation (RO) observations in numerical weather prediction using observing system simulation experiments (OSSE). Three experiments are carried out: a Control experiment with no RO observations and two experiments with 14,000 RO profiles per day—one with only polar orbiting satellites which provides a high concentration of profiles in the extratropics but has gaps in spatial coverage every 6 hr, and a second with a mix of polar and equatorial orbits with RO profiles that are more concentrated in the tropics and a more uniform local time coverage. Both experiments with RO included show a marked increase in forecast accuracy compared to the Control. The mixed orbit case shows improvement in forecast accuracy compared to the polar‐only orbit case in the tropics. The OSSE framework allows calculation of observation impacts on background and analysis errors, so that the contribution of information from retained RO information can be compared with the contribution of new information from observations ingested in the current cycle. Impacts of RO in the two sampling experiments are also explored using a forecast sensitivity observation impact tool for 24‐hr forecasts, and by comparing the skill of medium‐range forecasts. In stratospheric midlatitudes, the RO observational information is well retained, so that more frequent sampling is not required. However, the Tropics are found to be particularly sensitive to RO data in both the stratosphere and troposphere, and relatively uniform coverage is advantageous.