This paper describes a complete end-to-end Doppler centroid estimation scheme, which determines the fractional PRF part of the Doppler centroid. It also resolves the Doppler ambiguity. Experiments show that the scheme works successfully over various terrain types, including land, water, and ice, and that it requires only a modest amount of SAR data to perform reliably. The proposed scheme has an added advantage that it is directly applicable to RADARSAT and ENVISAT ScanSAR data. The scheme uses two complementary Doppler estimation algorithms, both utilizing the phase information embedded in the radar signal. In each algorithm, upper and lower parts of the available bandwidth of the received signal are extracted to form two range looks. The first algorithm, called multilook cross correlation (MLCC), computes the average cross correlation coefficient between adjacent azimuth samples for each of the two looks and then takes the difference between the angles of the two coefficients. The Doppler ambiguity is determined from the angle difference. The fractional pulse repetition frequency (PRF) part is also determined from the cross correlation coefficients, in the second algorithm, called multilook beat frequency (MLBF), the two looks are multiplied together to generate a beat signal. The beat frequency is then estimated and the Doppler ambiguity determined from the beat frequency. The MLCC algorithm performs better with low contrast scenes while the MLBF works better with high contrast ones. Although each algorithm works well on its own with sufficient averaging, it is also possible to use quality measures to select the best result from either algorithm. In this way, scenes of different content or contrast can be handled reliably. This paper presents the analysis of the two algorithms, explaining why their performance is affected by scene contrast, which is confirmed by experimental results with ERS-1 and JERS-1 data.
