Abstract
Error variance in a horizontal divergence field due to random error in raw radial velocity data was numerically estimated in order to apply the “floating boundary condition” concept developed by Chong and Testud (1983) to the adjustment of vertical wind fields derived from conical-scan-based dual-Doppler radar observation. A filter for interpolating raw radial velocity data onto common grids consisted of a combination of distance-weighted spatial averaging and a Cressman weighting function. Two cases, —shallow and deep—, were considered with error variance and gain estimated for both, using three influence volumes of sphere and oblate spheroids. Results without vertical shear showed for the two cases that the filter retrieved original wind fields well regardless of the shapes of influence volume considered, and that the distortion of wind fields through filtering was negligible for the horizontal scale of meteorological interests to be observed by dual-Doppler radar synthesis. For such a scale, the error variance was considered constant, and was almost equal to that in noise only, in which random noise alone accounted for simulated Doppler velocities. Based on results for noise only, a simple way to estimate error variance in horizontal divergence in terms of rms of random error in raw radial velocity data was presented for different baseline lengths. These estimates may be used for most vertical wind adjustment by floating boundary condition because the presence of vertical shear would not considerably alter estimates.
