Numerical wave modelling of Norwegian coastal areas is challenging due to the drastically varying bathymetry, irregular coastline, and large domains of interest. The widely used phase-averaging models are limited by such bathymetric and topographic conditions. Phase-resolving models provide higher accuracy regarding strong nonlinear wave transformations with the trade-off of a higher computational cost. This study provides a combined phase-averaging and phase-resolving modelling approach to analyse the wave conditions for on-shore aquaculture facilities at the site candidates Fiskenes and Breivik, Andøya, Norway. The phase-averaging spectral model SWAN is used for the offshore sea state analysis based on the offshore hind-cast data. The analysis is performed on cascade nested grids, with increasing accuracy closer to the proposed harbour locations. A novel interpolation algorithm is proposed to provide comprehensive sea state information with every offshore wave directions without requiring additional simulations. The critical wave conditions are identified from SWAN and used as input to the open-source phase-resolving fully nonlinear potential flow model REEF3D::FNPF. Four scenarios are investigated and densely spaced wave gauges in the entire computational domain provide the distribution of significant wave height. The combined and cascade simulation approach helps to achieve a balance between computational efficiency and accuracy for large-scale marine environment assessment. The results show detailed wave statistics in the entire area of relevance near both harbours and provide a quantitative reference for both the site choice and the harbour design.