Wave energy resource assessment

The use of satellite altimeter data for spatial mapping of the wave resource is examined.A new algorithm for estimating wave period from altimeter data is developed andvalidated, which enables estimates of wave energy converter (WEC) power to bederived. Maps of the long-term mean WEC power from altimeter data are of a higherspatial resolution than is available from global wave model data. They can be used foridentifying promising wave energy locations along particular stretches of coastline,before a detailed study using nearshore models is undertaken.The accuracy of estimates of WEC power from wave model data is considered. Withoutcalibration estimates of the mean WEC power from model data can be biased of theorder of 10-20%. The calibration of wave model data is complicated by non-lineardependence of model parameters on multiple factors, and seasonal and interannualchanges in biases. After calibration the accuracy in the estimate of the historic powerproduction at a site is of the order of 5%, but the changing biases make it difficult tospecify the accuracy more precisely.The accuracy of predictions of the future energy yield from a WEC is limited by theaccuracy of the historic data and the variability in the resource. The variability in 5, 10and 20 year mean power levels is studied for an area in the north of Scotland, andshown to be greater than if annual power anomalies were uncorrelated noise. Thesensitivity of WEC power production to climate change is also examined, and it isshown that the change in wave climate over the life time of a wave farm is likely to besmall in comparison to the natural level of variability. It is shown that despite theuncertainty related to variability in the wave climate, improvements in the accuracy ofhistoric data will improve the accuracy of predictions of future WEC yield.The topic of extreme wave analysis is also considered. A comparison of estimators forthe generalised Pareto distribution (GPD) is presented. It is recommended that theLikelihood-Moment estimator should be used in preference to other estimators for theGPD. The use of seasonal models for extremes is also considered. In contrast toassertions made in previous studies, it is demonstrated that non-seasonal models have alower bias and variance than models which analyse the data in separate seasons