Algorithms for sea surface wind parameter extraction from x-band shipborne nautical radar images

摘要

In this thesis, research for improving sea surface wind parameter extraction fromudshipborne X-band marine radar images is presented. First, the curve-fitting-basedudwind algorithms are investigated. To exclude the rain cases and low-backscatter images,uda data quality control process is designed. Then, a dual-curve-fitting techniqueudis proposed to enhance the wind retrieval performance under low sea states. Thisudmodified curve-fitting-based wind algorithm is tested using radar images and shipborneudanemometer data collected on the east coast of Canada. It is shown that theuddual-curve-fitting algorithm produces improvements in the mean differences betweenudthe radar and the anemometer results for wind direction and speed of about 5.7◦ andud0.3 m/s, respectively, under sea states with significant wave height lower than 2.30 m.udSecondly, the intensity-level-selection- (ILS-) based wind algorithms are studied. Anudadditional process is implemented for the ILS-based method to improve the accuracyudof wind measurements, including the recognition of blockages and islands in the temporallyudintegrated radar images. Moreover, a harmonic function that is least-squaresudfitted to the selected range distances vector as a function of antenna look directionudis applied. This modified ILS-based wind algorithm is applied to the same radaruddata. Compared with the original ILS-based algorithm, the modified one reducesudthe standard deviation (STD) of wind direction and speed by about 4◦ and 0.2 m/s,udrespectively. Also, the above mentioned two modified methods (dual-curve-fittingbasedudand modified ILS-based) are compared. Finally, X-band radar signatures ofudrain cells are elaborately described both in time and frequency domains. It is seenudthat the rain-contaminated image pixels are more uniformly bright than the waveudechoes and radar retrieved wind results are thus overestimated. This property ofud“uniformly bright” is used to identify the portions that are more affected by rain.udTo mitigate the effects of rain on wind retrieval from X-band radar images, a noveludtexture-analysis-based data filtering process is presented and tested. By removing the data in the directions more affected by rain, significant improvements can be seenudfrom the radar-derived wind results using both of the two wind algorithms.