Improving Nearshore Wave Guidance at the North American Great Lakes: Explicit and Fully Implicit integration of the Wave Action Balance Equation

The National Centers for Environmental Prediction (NCEP) of the US National Weather Service provides wave guidance to the North America Great Lakes since 2004, using an implementation of the state-of- the-art numerical wind-wave model WAVEWATCH III that covers its 5 major lake basins: Superior, Michigan, Erie/St.Clair, Huron and Ontario. The local conditions, with mid-latitude cyclones, Arctic air mass intrusions, and organized intense convective systems, in association with a complex coastal geometry, provide a spectacular opportunity for challenging the skill of wave models. The impact of upgrading NCEP’s Great Lakes wave model system (GLW) to improving nearshore wave guidance in the region will be discussed, as a consequence of adding an unstructured spatial grid.

The new Great Lakes unstructured grid brought effectively a ten-fold increase in nearshore resolution for NCEP’s operational system, while keeping compute-resource usage viable. Explicit and implicit schemes were tested as part of the investigation on optimal configurations for operational applications, given the available HPC resources. An initial implementation was made using an explicit spatial propagation scheme, given an initial target of 250m resolution near the coast. The use of implicit schemes in that configuration did not exhibit computational effort advantages. As the wave model focus approaches the shore for the development of more accurate products for beach hazard prediction, however, the CFL restriction of the explicit scheme becomes an overpowering limitation.

Implicit schemes do not obey the CFL restriction. With the recent inclusion of implicit schemes in the WAVEWATCH III package, it is anticipated that near-future upgrades to the GLW system will jump over the severe time step constraint imposed by explicit methods and reach much more efficient integration. The current work presented here investigates the consequences of applying implicit time stepping schemes for very-high nearshore resolution in the GLW system. Our results show that the skill scores and the efficiency of the implicit scheme were consistent with those of the explicit approach for resolutions up to 250m near the coast. However, the grid generation process becomes a significant part of the efficiency of the scheme and the time spend to implement may be a tradeoff relative to the higher flexibility of implicit schemes.

Future developments of the GLW, however, will require increasing the spatial resolution nearshore to the order of 10m at the coast. Our investigation suggests that in order to attend such challenging future demands of that system, the implicit schemes in WAVEWATCH III provide an adequate framework. Results indeed indicate that veering model development towards using implicit spatial propagation schemes provides the GLW an optimal framework for attaining very-high resolution wave guidance near the coast, whilst keeping data with high-quality skill scores and high computational efficiency.