The quality assurance (QA) algorithm infers the quality of scatterometer retrieved winds by evaluating the self-consistency of the selected wind flow. This is done in two ways:

• First, the selected wind is compared to a low order Karhunen Loeve (KL) model fit on a region-by-region basis. Individual cells are flagged where the directional or vector error exceeds thresholds. The rms vector error and number of flagged cells are computed for each region.
• Second, the directions for each region are binned and the resulting histogram is inspected for multimodalities.

Each wind vector cell is given a level QA bit flag defined in the adjacent table. Cells flagged with both the region ASE flag and the ASE cell flag have a high likelhood of indicating the position of ambiguity selection errors. The QA flag can be viewed as an integer number between 0 and 15 indicating increasing uncertainty of the selected wind flow. Because the QA algorithm is effective in locating unnatural breaks in the wind flow, areas with realistic fine-scale wind features (such as storms and fronts) may be flagged while ambiguity selection errors resulting in consistant wind flow may not.

[q3,q2,q1,q0] QA Flag [x,x,x,1] Noisy Vector Cell Flag [x,x,1,x] ASE Cell Flag [0,0,x,x] Region "Good" Rating [0,1,x,x] Region "Fair" Rating [1,0,x,x] Region "Poor" Rating [1,1,x,x] Region ASE Flag