Satellite and In-situ Observations: Complementarity and Limitations
The last lecture introduced the many platforms that observe the ocean. This one, again with Frederic Bonou, asks a sharper question: if we have all these systems, why do we need so many? The answer is that no single observing system can fully capture the ocean – and understanding each one’s strengths and limits is what reveals why they belong together.
The lecture compares the two families head to head:
- What satellites do best – a global, synoptic view of the ocean surface. In a single image, a satellite can show sea surface temperature across every basin, repeated consistently over time. No in-situ network could achieve that coverage
- Where satellites fall short – they mostly see only the surface “skin,” so they can’t measure temperature, salinity, or currents hundreds of metres down; some sensors are blocked by cloud, rain, or sea ice; and their products must be validated against direct measurements
- What in-situ observations do best – direct, highly accurate measurements within the ocean, throughout the water column from surface to deep. This is the information satellites simply cannot provide – and it’s what calibrates and validates both satellite products and models
- Where in-situ observations fall short – they sample only discrete points, not continuously; coverage is uneven and costly to maintain, thin in remote regions, marginal seas, and the deep ocean; and filling the gaps between measurements requires interpolation
- Why both are needed – satellites for broad surface coverage, in-situ for depth and accuracy. Combined with numerical models, the two together give a far more complete and reliable picture of the ocean than either alone
The takeaway is the heart of the week: satellite and in-situ observations don’t compete – they complement each other. Their strengths and weaknesses are mirror images, which is exactly why an integrated observing system, feeding data assimilation, is what modern forecasting relies on. The next lecture opens up that assimilation step – how observations and models are actually combined.
