Abstract
Measured wave spectra are crucial for direct characterisation of wave resource, predicting loads on offshore structures, and driving models of nearshore waves, wave-current interactions and sediment dynamics. Although single-point devices such as wave buoys are only able to supply the ºfirst fiveº directional moments in each frequency bin, an accurate reconstruction of the full directional spectrum may be obtained by imposing additional constraints. The most successful such technique for reproducing real sea data is the Maximum Entropy Method (MEM). However, computational complexity and perennial issues with convergence act as a barrier to its widespread implementation. This forces researchers to resort to cruder, single-parameter distributions such as Gaussian and cos2s, thereby neglecting half of the available directional information. By reformulating the underlying analytical equationss, we have produced a new numerical implementation of the MEM with greatly improved speed and convergence properties, allowing rapid evaluation of directional spectra from raw buoy data. We have applied this new open source tool to a year’s worth of buoy data gathered at the UK Wave Hub Test Site. The more detailed Maximum Entropy spectra exhibit less directional spreading than those derived from standard single-parameter techniques, and can display bimodal features. In particular, clear differences can be seen for lower wave frequencies that correspond to swell waves. In such cases, relying on simpler parametrised formulae can introduce systematic errors when reconstructing the parts of the spectrum most relevant for energy harvesting and storm prediction, which are avoided in this new streamlined treatment.