Assimilation of Atmospheric Radio Refractivity Using a Nonhydrostatic Adjoint Model

Abstract

Recently, a new approach to remote sensing of water vapor based on the Global Positioning System (GPS) has been proposed. Specifically, the bending of radio signals propagating from GPS satellites to a receiver on a low earth-orbiting satellite can be used to derive vertical profiles of atmospheric refractivity. Vertical profiles of temperature and water vapor can then be retrieved from the refractivity measurements. This is potentially a valuable data source for the meteorological community. However, before such measurements are used for operational numerical weather prediction, we need to assess the accuracy of the retrieved temperature and moisture fields and properly assimilate these observations into a numerical model. A 4D data assimilation system based on the adiabatic version of the Penn State-NCAR Mesoscale Model and its adjoint was developed. A series of observing system simulation experiments was then conducted to assess the impact of GPS-derived atmospheric refractivity data. Specifically, a 20-km simulation of a winter storm in March 1992 over the continental United States was used as the control experiment. Vertical profiles of atmospheric refractivity were extracted from the control simulation at selected temporal and spatial resolutions. The simulated GPS measurements were then assimilated into a 60-km MM5 using the 4D variational data assimilation approach. The results showed that the assimilation of atmospheric refractivity is very effective in recovering the vertical profiles of water vapor. The accuracy of the derived water vapor field is significantly better than that obtained through the traditional retrieval technique. The assimilation of atmospheric refractivity is also shown to provide useful temperature information. The data assimilation results are relatively insensitive to the random errors added to the simulated refractivity observations. However, they are very sensitive to the spatial resolution of the observations. A spectral analysis shows that the moisture field has more small-scale variation and is more sensitive to the resolution of the atmospheric refractivity observations than the temperature. When refractivity observations are available on a coarser resolution, assimilation of individual observations produces better results than assimilation of the interpolated observations on the model grid. Assimilation of the refractivity observations averaged along a distance of about 240 km, which represents the characteristic scale of the GPS refractivity measurement, still produces reasonably good results for the retrieval of temperature and moisture fields.