Numerical wave models are often used to hindcast wave conditions and predict the theoretical energy production from wave energy conversion (WEC) devices. It is widely acknowledged that numerical model suffer from bias's and uncertainties which ultimately affect the final predictions of WEC power. In this case study, a Simulating WAves Nearshore (SWAN) hindcast, based on the ECMWF wave and FNMOC wind boundary conditions, is used to predict sea states off the Canadian west coast and the triple collocation technique is applied to quantify the model result bias's, and systematic and random errors. To analyze the error and calibrate the results from the hindcast, two in-situ collocated wave measurement devices are deployed; A TriAxys wave measurement buoy and a Nortek AWAC. The triple collocation technique is used to compare the significant wave height and energy period parameters over a threemonth period, from October to December 2014. The triple collocation technique assumes linear relationship between the measured value and true value, and outputs the bias, calibration slope and the measurement random error. This study implements two previously utilized calibration regimes, a single value and monthly calibration regime, as well as presenting two novel methods to improve the impact of the calibration; a bivariate calibration and a spectral calibration. The two standard calibration techniques resulted in negligible improvements in data correlation. The spectral method suffered from high computational cost and only 3.4% improvement in significant wave height correlation. The bivariate calibration regime, following the International Electrotechnical Commission (IEC) wave resource histogram parameters, resulted in 5% and 26% improvements in the significant wave height and energy period correlations respectively. Calibration of the SWAN hindcast reduced the gross wave energy transport values by 900 MWh, yet the final WEC production estimates only varied by 1.5 MWh but greatly improved their time-series correlation. It is shown that the triple collocation technique, under the bivariate distribution regime, provides a more realistic presentation of future WEC power production and eliminates known short-comings of numerical model outputs. (C) 2016 Elsevier Ltd. All rights reserved.