Simulation of Wind Gradient Errors in NROSS (Navy Remote Ocean Sensing System) Radar Scatterometer Data in a Simplified Geometry

摘要

A set of simplified case studies is used to illuminate the effect of spatial gradients in normalized radar cross section (NRCS) on the accuracy of NRCS at NROSS (Navy Remote Ocean Sensing System) scatterometer (NSCAT) cell centroids, and at wind vector retrieval grid points displaced from the centroid. Values of linear and quadratic variables s(x,y) are estimated at the centroid and at a point displaced from the centroid of a parallelogram cell, using first the mean value (S) over the cell (a nearest neighbor or binning approach) and then a bilinear interpolation estimator Si(x,y). Isoparametric finite elements are used to take cell shape into account in the interpolation. At cell centroids with linear fields s(x,y), errors are negligible using either (S) or Si(x,y). At cell centroids with nonlinear fields s(x,y), errors using (S) and Si(x,y) both approach 0.5% (with interpolation errors being slightly less) at spatial gradients of 0.1 dB/km. At displaced points with linear fields s(x,y), errors using interpolation Si(x,y) are negligible at all gradients, but error using cell means (S) exceeds 6% at gradients of 0.1 dB/km. At displaced points with nonlinear fields s(x,y), error using interpolation Si(x,y) approaches 2%, while error using cell means (S) exceeds 7%, near spatial gradients of 0.1 dB/km. The simulated errors associated with binning are large, and interpolation yields much better accuracy. Additional simulations using realistic wind fields and NSCAT Doppler cell patterns should be carried out to further quantify this component of the error budget.