Every day, satellites capture valuable images of the ocean at many spatial scales. While fine scales play a major role in many ocean phenomena, the accuracy of satellite sensors is crucial for tracking them. A team from the University of California has specifically investigated the reliability of the data by comparing two generations of space sensors, with the aim of detecting any technical artefacts.
by Laurie Henry
Cover photo: MODIS satellite. © Public Domain / NASA
The ocean regulates a significant portion of the Earth’s climate by redistributing heat, carbon, and nutrients through a multitude of surface currents and vertical movements. Phenomena known as submesoscale phenomena—ranging from a few hundred metres to tens of kilometres—are particularly important. These fine structures directly influence ocean circulation, phytoplankton distribution and exchanges with the atmosphere.
Capturing these small scales of ocean circulation on a global scale remains a challenge. Satellites are the only tools capable of observing all of the oceans every day and in their entirety. However, their ability to accurately reproduce the finest details is limited: spatial resolution and instrumental noise can distort measurements, particularly those of surface temperature and chlorophyll concentration. A team from the University of California has specifically investigated this reliability by comparing two generations of space sensors.
The major role of small-scale currents
At the boundary between ocean turbulence and global circulation, submesoscale currents play a crucial role in ocean dynamics, particularly because they promote vertical exchange. These rapid movements transport heat, carbon dioxide and nutrients between the surface and deeper layers. They also directly influence the distribution of phytoplankton, on which a large part of the marine food chain depends.
Despite their importance, these structures remain difficult to detect and characterise on a large scale. Observation satellites remain the only tools capable of tracking these phenomena on a global scale. However, their spatial resolution, which is often comparable to that of the structures they seek to measure, makes observations difficult. The slightest noise in the data – an instrumental fluctuation, calibration inaccuracy or atmospheric effect – can appear as a real physical variation. By comparing data from the MODIS Aqua sensor (whose images have a resolution of 1.5 km for chlorophyll and 1 km for temperature) with those from Landsat 8/9 (whose images have a resolution of 30 metres for chlorophyll and 100 metres for temperature), the authors of the study were able to test how resolution influences the measurement of sub-mesoscale variations, particularly for two fundamental tracers: surface temperature and chlorophyll.
Adjust resolution and frequency to detect small scales
The results are clear: MODIS shows much more pronounced changes than Landsat in the data. This does not mean that MODIS detects details better: in fact, this impression of greater complexity stems from a technical flaw. MODIS introduces ‘noise’ into its images — errors that are invisible at first glance but disrupt the data on a small scale. Each pixel does not only reflect the actual state of the ocean, it is also influenced by inaccuracies related to the instrument itself.
To ensure that this difference was not simply due to resolution, the researchers simulated Landsat images at the same scale as those from MODIS. This test, known as L@M, showed that when the spatial quality of Landsat images is deliberately reduced, they remain more reliable than those from MODIS. This proves that the problem is not solely due to pixel size, but rather to the level of measurement accuracy.
Chlorophyll concentration (a–c, g–i) and surface temperature (d–f, j–l) observed by Landsat 8 (left column), MODIS Aqua (centre) and Landsat resampled at MODIS resolution (L@M, right), off Point Conception (California) on 30/11/2017. The first two lines show the entire area, the bottom lines show a zoom. The dotted red line delimits the spectral analysis area and the cyan indicates the enlarged sub-area. © Luke Carberry et al., 2025
Some of the ‘turbulence’ revealed by MODIS therefore appears not to exist and to be due solely to an instrumental error. This technical artefact gives the illusion of a more turbulent or complex ocean than it really is, which can distort the analysis of currents, biological production, or climate exchanges.
Measure technical limitations before interpreting images
This bias then compromises the analysis of links between ocean dynamics (such as water temperature) and biological activity (such as chlorophyll concentration). Surface currents influence the distribution of marine life, particularly phytoplankton, which forms the basis of the entire ocean food chain. These physical-biological links are particularly visible at small scales where eddies, filaments and other marine fronts form (between 1 and 30 kilometres).
To study this interaction, researchers compared how each satellite reports differences in temperature and chlorophyll at these scales. Theoretically, chlorophyll is expected to show more detail than temperature because it reacts more quickly to water movements and localised nutrient inputs. This is what Landsat data reveals, showing smaller and more numerous structures in chlorophyll images than in temperature images.
Landast 9. © Public Domain / NASA
This difference disappears with MODIS images. The two parameters appear to vary almost identically, as if biology did not react differently to physical conditions. This result, which does not reflect the reality of the ocean, seems to be related to a measurement error. Noise in MODIS images blurs contrasts and masks temperature-phytoplankton links, drowning the real biological signal in a technical artefact linked to instrumental error.
Satellite data must therefore be interpreted with caution and their technical limitations taken into account before interpreting the images.
Technology at the service of scientific objectives
Comparative analysis between MODIS and Landsat shows that a satellite’s ability to observe the ocean in detail does not depend solely on the size of its pixels. The stability and reliability of measurements, in other words the signal-to-noise ratio, seems to be just as important, if not more so. A satellite such as MODIS may have global daily coverage, but if it introduces too much noise into its data, it loses its ability to reliably detect fine structures.
Conversely, Landsat, with much smaller pixels and higher noise per pixel, manages to produce more consistent data when its measurements are grouped together. This provides an accurate picture of variations at sub-mesoscale scales without being distorted by sensor errors. This approach confirms a strategy that is undoubtedly effective: using high-spatial-resolution satellites even if they cover a smaller area or have lower frequency.
Ocean chlorophyll represented with a perceptible colour scale, centred on the long-term average to highlight local variations, MODIS data. © NASA’s Goddard Space Flight Center
These findings have practical implications for future ocean observation missions, such as the PACE satellite, which carries hyperspectral instruments to better measure water colour. For these new sensors to be truly useful in studying small-scale ocean processes, they must strike a good balance between spatial resolution, signal quality and observation frequency. Satellite designers will need to take into account that, when studying fine currents or local biological responses, a poor signal-to-noise ratio can negate the expected gains from high resolution.
Another avenue being explored is to rethink observation strategies using constellations of high-spatial-resolution satellites. Even if each of these satellites observes less frequently, their combination could provide sufficient coverage while maintaining a measurement quality suitable for studying sub-mesoscale structures. This approach requires coordination and data processing efforts, but it better meets the accuracy requirements identified in the study.
Finally, the results highlight a key point for the scientific community and space mission operators: any interpretation based on satellite images must take into account the instrumental limitations of the sensors. Regardless of the technical choice, the requirement remains the same: to understand the ocean at small scales, clear, reliable data that is appropriate for the structures being observed is needed.
Source : Carberry, L., Siegel, D. A., & Nidzieko, N. J. (2025). “Effect of satellite spatial resolution on submesoscale variance in ocean color and temperature”. Geophysical Research Letters, 52, e2024GL114266.