01. Current Research and Work

My work at Fisheries and Oceans Canada is centered around building ocean circulation models, primarily to provide data for both e-navigational products and drift prediction tools for emergency and spill response. I mostly work on models of the Bay of Fundy (particularly around Saint John, New Brunswick), though I work with colleagues on other Canadian waters as well.

The Bay of Fundy is famous for having the highest tides in the world, with the tidal range reaching ~16m in the uppermost parts of the bay. The city of Saint John, NB is located about halfway up the Bay at the mouth of the Wolastoq (Saint John River), and the tidal range is only (!) about 8 metres. The Wolastq is the largest river emptying into the Bay of Fundy, with a winding course and several sills near the outlet at the harbour. The interaction of the river and tides make for a complex circulation in the harbour: one of the sills in the river is high enough (ie blocks enough river flow) that at high tide, the water in the river flows upstream and inland, rather than downstream and into the ocean. This unusual place is called Reversing Falls, and this regular oscillation of flow is part of what makes the circulation in the harbour so complex.

I use the NEMO ocean model at high horizontal resolution to run fully baroclinic circulation models at harbour scales. For Saint John Harbour, I use a two level downscaling system, with the first level covering the entire Bay of Fundy (dx ~ 500m) and the second level covering the port of Saint John and its approach (dx ~ 100m). I first ran a 7 year hindcast from 2016 to 2023, and the system is designed to seamlessly transition to a daily real-time and forecasting system. Every day, the hindcast is extended by a single day, and a two day forecast is launched. This forecasting system is currently being run daily with "best-effort" support, and is being used to develop products that will eventually be used by pilots, fishers, and other harbour users to inform their activities and help them navigate a complex harbour.

Developing ocean models requires tools to evaluate how accurately the model can reproduce observations. To do this, I have developed sets of scripts that can efficiently and robustly compare model output on a regular grid to observations gathered from a variety of types of instruments, including tide gauges, ADCPs and current metres, CTDs, and surface buoys. I've used these scripts to help evaluate several coastal, regional, and nearshore systems developed at DFO and ECCC, and am currently expanding the tools to be able to evaluate forecasts as well as hindcasts.

Modelling work typically requires intensive computer resources, and NEMO is a complex and computationally heavy model. As a result of doing key work on computer migrations on government HPC machines (as well as previous work done during my PhD), I have become skilled at making ocean models compile properly and run efficiently on anywhere from 10 to ~800 cores on a variety of HPC systems.