A two-dimensional variational (2DVAR) system is developed for the analysis of near-surface moisture from radar refractivity-related phase change observations. Refractivity can be calculated from the phase change between two ground targets aligned along the radar beams. Being most sensitive to atmospheric moisture content, refractivity observations can provide high-resolution information on the often highly spatially varying low-level moisture. Thus, the assimilation of the electromagnetic wave phase change fields as measured by radar is expected to improve convective initiation and quantitative precipitation forecasting.

In this study, simulated radar refractivity-related phase change data are created using a radar simulator from realistic high-resolution model simulation data from a convective initiation case of the 2002 International H2O Project (IHOP_2002) field experiment. The simulated observations are then analyzed. It is found that the 2DVAR system, without the help of any additional information, is incapable of dealing with 2-pi phase-wrapped observations. When no phase wrapping is present in the data, the 2DVAR produces excellent analyses that agree with the truth well, and this is true when the background error covariance of proper scale is modeled using either spatially isotropic or anisotropic filter. When phase wrapping is present, a signal processing procedure for phase-unwrapped is applied and the resultant, somewhat spatially smoothed observations, are analyzed; rather accurate moisture analysis is still obtained. The moisture field contains many small scale structures associated boundary convective eddies and rolls found on both sides of a dry line. Sensitivity experiments show that the analysis system is robust to random observation error, error associated with ground target position uncertainties between the range gates, and error introduced by the phase wrapping procedure.