Satellite borne radar scatterometers provide frequent estimates of near surface wind vectors over the Earth's oceans. However in the polar oceans, the presence of sea ice in or near the measurement footprint can adversely a ect scatterometer measurements resulting in inaccurate wind estimates. Currently, such ice contamination is mitigated by discarding measurements within 50 km of detected sea ice. This approach is imperfect and causes loss of coverage. This thesis presents a new algorithm which detects ice-contaminated measurements based on a metric called the Ice Contribution Ratio (ICR) which measures the spatial ice contribution for each measurement. The ICR calculation is made for each measurement using a spatial ice probability map which is determined using Bayesian probability theory. Determined by simulation, the ICR processing thresholds the ICR for each measurement depending on local wind, ice backscatter, and cross-track location. ICR processing retrieves winds at a distance of 22.5 km from the ice edge on average, while ensuring wind accuracy. Retrieved wind distributions using ICR processing more closely resembles uncontaminated wind distributions than winds retrieved using previous methods. The algorithm is applied to QuikSCAT in this thesis but could be applied to other scatterometers such as the Oceansat-2 scatterometer.