What Are Upstream Data Sources in Ocean Forecasting?
This opening lecture, with Nathanaël Dossa, introduces the foundation of the whole module: upstream data sources – the observations and external datasets a numerical model needs before it can produce a forecast.
The key definition to anchor the week: in operational ocean forecasting, upstream data are the observations and external datasets required to initialize, force, and validate numerical ocean models. They describe the state of the ocean and its environment before a forecast can be produced. The lecture works through the main types, one by one:
- Satellite observations – continuous, near-global coverage of the ocean surface, though they cannot see beneath it
- In-situ observations – Argo floats, moorings, gliders, research vessels, drifting buoys, and coastal stations measuring temperature, salinity, currents, and more through the water column. Together with satellites, they give a comprehensive picture of the ocean
- Bathymetry – the depth and shape of the ocean floor, which controls how currents, tides, and waves behave
- Atmospheric forcing – the winds, heat fluxes, pressure, precipitation, evaporation, and solar radiation that drive the ocean from above
- River inputs – freshwater, sediments, and nutrients that rivers deliver to the coast, reshaping salinity, density, and circulation
- Parent model outputs – the boundary and initial conditions a global model provides to a higher-resolution regional model, through nesting
- Climatological datasets – long-term averages used to initialize models where observations are scarce, and to evaluate performance and anomalies
The lecture closes on why all this matters, with a principle that runs through the entire week: accurate forecasts begin with accurate data. However sophisticated a model is, its predictions are only as reliable as the information used to initialize and drive it. Observations tell us what the ocean looks like today; atmospheric forcing explains how the atmosphere influences it; the model predicts how it will evolve; and data assimilation combines them all. The next lecture asks where these observations actually come from – the global data providers.
