Oil Thickness Estimation Using Single- and Dual-Frequency Maximum-Likelihood Approach

In this paper, we present Maximum Likelihood single- and dual-frequency estimators for oil spill thickness estimation. The estimators use Minimum-Euclidean distance algorithm, in pre-defined 1-D or 2-D constellation sets, on simulated reflectivities to estimate the thickness of the oil on top of sea surface. Results show that the performance of the estimator is dependent on the radar frequency used and on the real thickness level. The steep slope of the reflectivity at some frequencies allows accurate estimation in some thickness ranges but it can cause significant errors in other ranges due to its periodicity. The monotonic behavior of the reflectivity at other frequencies lead to a less accurate estimation but smaller error range. Performance analysis of the dual-frequency estimator shows that it overcomes the drawbacks of each single-frequency estimator by taking the advantages of the monotonicity at the first frequency and the steepness of the slope of the reflectivity at the second frequency.