The sea clutter phenomenon is investigated from a different perspective by using the finite element domain decomposition (FEDD) method, which is a full-wave numerical method based on the decomposition of the problem into sub-problems with the help of the locally-conformal perfectly matched layer (LC-PML) approach. The numerical model developed in this work provides the means to investigate the sea clutter phenomenon by a full-wave Maxwell solver, although the electrical size of computational domain is formidably large. Monte Carlo simulations are performed to obtain some statistical properties of reflectivity. The results are compared with those obtained from existing empirical models and measured data for different polarizations, frequencies, grazing angles and wind speeds (or sea states). It is numerically shown that the probability distribution of the clutter signal power appears to follow the gamma distribution. Furthermore, the concept of ergodicity, which means that the statistical results obtained by ensemble averaging in a single cell are equivalent to those obtained by spatial averaging over a set of consecutive range cells in a radar system, is examined. It is observed from the numerical results that the behavior of the sea clutter random field is consistent with the ergodicity hypothesis. Some conclusions are drawn about the statistical behavior of the sea surface reflectivity. (C) 2021 Published by Elsevier B.V.