The design of coherent detectors for targets in sea clutter presents particular challenges. The Doppler spectra of sea clutter are observed to be both time-varying and range-varying so that when data from a large number of surrounding cells are used to estimate the characteristics of the data in the cell under test, the results can be unreliable. Attempts to use smaller numbers of possibly more representative cells can result in numerical stability problems. In this study the authors extend earlier work on this subject and assess a number of traditional and model-based techniques to see how well false alarms can be controlled while using data from only a small number of surrounding cells. The authors also compare the resulting detection performance when the detection thresholds are adjusted to give a constant false alarm rate at all Doppler frequencies.