A satellite can show where Arctic sea ice is, how it moves, and how it changes over time. But it cannot simply look down and read off every property of the ice. It measures radiation or a returning radar pulse. Scientists then use a relationship between that signal and the ice to estimate concentration, type, or thickness.

Fieldwork supplies measurements to build and test those relationships. It also reveals things that are difficult to recover from orbit, such as salt inside the ice or heat arriving from the water below. The question is not whether satellites or fieldwork are better. It is how to use each for the part the other cannot do well.

What field measurements are for

Think about a satellite map that reports ice thickness. Before trusting a value on that map, we need to answer two different questions:

• How was the value calculated? Measurements on the ice help connect the satellite signal to an actual ice property. Building or adjusting that relationship is calibration.
• Does the calculation work on new observations? Comparing the estimates with independent field measurements is validation.

The independent measurements are often called ground truth. That does not mean they are perfect. A drill measurement has uncertainty, and one hole may not represent the whole area in a pixel.

Fieldwork also lets us follow a process instead of comparing occasional snapshots. A buoy can record a temperature profile through the same floe for months, showing when and where ice grows or melts. Instruments in the ocean can measure the heat supplied from below. An ice core can show how salinity changes with depth.

These measurements connect back to the Arctic as a coupled system: changes in the atmosphere, snow, ice, and ocean affect one another. Measuring them together helps explain why the ice changed, not just that it changed.

Working on moving ice

A floe is a piece of floating sea ice. A camp on a floe moves with the ice rather than staying at one geographic location. That is useful when the aim is to follow the same ice through time.

There are several ways to work from it:

• Drifting camps put instruments and temporary shelters directly on the floe.
• Icebreakers provide laboratories, supplies, transport, and a base beside the field site.
• Buoys stay behind and transmit measurements after the people leave.

The MOSAiC expedition used the icebreaker Polarstern as a central observatory during 2019–2020. Researchers established an ice camp and a surrounding network of instruments, then followed the changing ice through the seasons. Measuring the atmosphere, snow, ice, ocean, and biology together made this much more informative than a short visit.[2]

A drifting station has limits. A lead, a crack or channel of open water in the ice, can open through the camp and separate people from equipment. Converging ice can build a pressure ridge, a pile of broken ice, through a measurement site. Cold, darkness, bad weather, and wildlife hazards make work slower and restrict access. The easiest place to measure is not necessarily the place that represents the ice well.

Why one hole cannot represent a whole pixel

A drill hole may be only $5\,\mathrm{cm}$ across. A passive-microwave pixel may be $25\,\mathrm{km}$ across. Between those scales are snow-probe lines, electromagnetic surveys, and finer-resolution radar images.

From a drill hole to a satellite pixel

• Field measurement
• Satellite pixel

The intermediate survey lengths are illustrative, not fixed instrument limits. The important difference is what each measurement represents: one spot, a line across the ice, or an area containing many different surfaces.

Suppose a pixel contains smooth first-year ice, a pressure ridge, deep snowdrifts, and a narrow lead. A hole in the smooth ice tells us the thickness there. It does not tell us the average thickness of that mixture.

Could we just drill more holes?

Take a $25\,\mathrm{km}\times25\,\mathrm{km}$ pixel and sample it on a grid with one hole every $100\,\mathrm{m}$. Using 250 sampling positions along each side gives

$$N=\left(\frac{25{,}000}{100}\right)^2=62{,}500\text{ holes}.$$

At ten minutes per hole and eight working hours per day,

$$\text{working days} =\frac{62{,}500\times10}{8\times60} \approx1302.$$

That is about 3.6 years of working every day, before travel, weather delays, or equipment problems. The ice would move and change long before the survey finished.

Instead, scientists choose samples that cover the different conditions within the area. A transect is a line along which measurements are taken. Transects across smooth ice, ridges, and snowdrifts reveal variation that a single convenient location would miss.

The aim is not simply to collect many measurements. It is to collect measurements that represent the ice being studied.

Why buoys do not solve the coverage problem

Buoys are good at repeated measurements through time, but they are far apart. To see the scale of the problem, spread $N$ buoys evenly over an area $A$. Each buoy represents roughly $A/N$, so a rough spacing is

$$d\approx\sqrt{\frac{A}{N}}.$$

Here, $d$ is spacing in kilometres when $A$ is in square kilometres. For a winter ice area of $14\text{ million km}^2$, 100 buoys would be about 374 km apart. Even 1000 would be about 118 km apart.

More buoys still leave wide gaps

This is an idealized spacing calculation, not a map of an actual buoy network. It explains the trade-off: a buoy can record detailed changes at its location, while a satellite observes conditions across the basin.

Measuring thickness and snow

A drill hole gives a direct reference

An auger drills through the ice. A weighted tape with a hinged bar is lowered through the hole, then pulled upward until the bar catches the underside. Measuring from the top of the ice to that bar gives the ice thickness.

The same site can provide snow depth and freeboard, the height of the ice surface above the water. These measurements help check faster methods, but their small footprint still matters: a precise measurement at one spot can be a poor estimate of a large area's average.

An electromagnetic survey covers more ground

An electromagnetic (EM) instrument uses the contrast in electrical conductivity between sea ice and salty seawater. It generates a changing magnetic field. The conducting water produces a response that the instrument measures.

For comparable conditions, water closer to the instrument gives a stronger response. Water farther away gives a weaker response. This lets an instrument on a sled estimate the distance to the water while moving along a transect.

An EM sled senses how far away the conductive seawater is

Snow depth is measured separately before the sensor-to-water distance is converted to sea-ice thickness.

The catch is that distance to the water includes more than the ice. After accounting for the sensor's height above the snow, the remaining distance includes both snow and ice. The EM measurement alone does not cleanly separate them.

For example, if $30\,\mathrm{cm}$ of snow is left unaccounted for, it can be counted as part of the ice thickness. A snow-depth measurement is needed to separate the layers.

Snow needs a transect too

A snow probe is pushed down until it reaches the ice surface. Its penetration depth gives the snow depth, and the position is recorded with the reading.

Snow collects in hollows and drifts around ridges. One probe reading cannot describe that variation. Repeated measurements along a line give a much better picture of the snow over a floe or within a pixel.

Snow matters for satellite thickness estimates as well. Its depth and density affect the conversion from freeboard to thickness. In the approximate sensitivity example, a $10\,\mathrm{cm}$ snow-depth error produces about a $30\,\mathrm{cm}$ ice-thickness error. That factor is not universal; it depends on the conversion and the assumed densities.

Snow depth is difficult to retrieve remotely, but it is not correct to say there are no satellite products for it. ESA introduced an operational product combining CryoSat radar and ICESat-2 laser measurements in 2025. Snow remains a major source of uncertainty, so these estimates still need field checks.[1]

Reading a temperature profile

A thermistor string is a vertical chain of temperature sensors. Once frozen into place, it can record temperatures through the snow, ice, and water, often with sensors about $10\,\mathrm{cm}$ apart. Repeated profiles help track freezing and melting after the field team has left.

In a simple midwinter example, the top of the snow is at $-25^\circ\mathrm{C}$ and the water below the ice is at $-1.8^\circ\mathrm{C}$. The snow is $0.10\,\mathrm{m}$ thick and the ice is $1.50\,\mathrm{m}$ thick.

A midwinter temperature profile has a kink at the snow-ice boundary

Read the graph downward from the snow surface:

1. Temperature rises quickly through the thin snow layer.
2. At $0.10\,\mathrm{m}$ depth, the slope changes because the material changes from snow to ice.
3. Temperature then rises more gradually per metre through the ice, reaching the water temperature at $1.60\,\mathrm{m}$ depth.

The bottom is 1.60 m below the snow surface, but the ice thickness is 1.50 m. The other 0.10 m is snow.

Why is there a kink?

Snow is a better insulator than ice. In a steady one-dimensional model, the same heat must pass upward through both layers. Because snow conducts heat less readily, it needs a larger temperature change per metre to carry that heat.

Thermal conductivity, written $k$, measures how readily a material conducts heat. Using $k_s=0.3\,\mathrm{W\,m^{-1}\,K^{-1}}$ for snow and $k_i=2.0\,\mathrm{W\,m^{-1}\,K^{-1}}$ for ice gives a snow–ice interface temperature of about $-17.9^\circ\mathrm{C}$. Just 10 cm of snow accounts for about $7^\circ\mathrm{C}$ of the total $23.2^\circ\mathrm{C}$ temperature difference.

There is a small graph-reading trap here: temperature is horizontal and depth is vertical. A larger temperature change per metre makes the snow segment look flatter on the page, not more vertical.

The two straight segments describe an idealized steady winter profile. Real profiles change with weather, melt, flooding, and other processes. Sensors have finite spacing, so the boundaries are estimated rather than read with unlimited precision.

Measuring salt, albedo, and heat exchange

Salt inside the ice

An ice core preserves a vertical section of the ice. Researchers cut it into short sections, melt them separately, and measure the meltwater's electrical conductivity to estimate salinity. Keeping the sections separate gives a salinity profile rather than one average for the whole core.

Temperature can also be measured along the fresh core before it warms. Together, salinity and temperature help explain brine conditions, ice strength, and radar response. This connects to brine in growing sea ice.

Albedo at the surface

Albedo is the fraction of incoming sunlight reflected by a surface. An upward-looking sensor measures incoming shortwave radiation; a downward-looking sensor measures the radiation reflected upward from the surface. Using simultaneous readings,

$$\alpha=\frac{\text{reflected shortwave radiation}}{\text{incoming shortwave radiation}}.$$

Both readings use the same units, so albedo has no units. Measuring over snow, bare ice, and melt ponds helps explain why an average over a whole satellite pixel differs from a measurement at one spot.

The rest of the energy budget

Other instruments measure incoming and outgoing longwave radiation and the turbulent exchanges of sensible and latent heat. Temperature profiles and ocean measurements help estimate heat transferred through the ice and supplied from below.

These are the physical terms in the surface energy balance, now tied to measurements. A change in ice thickness is easier to explain when we also know where the energy went.

Connecting field measurements to a satellite map

A useful survey connects measurements at overlapping scales:

Illustrative lengths might be a 100 m drill line, a 1 km sled transect, a 10 km airborne survey, and a 25 km pixel. These are survey-design choices, not fixed resolutions for all instruments.

Each step increases coverage, but it does not remove uncertainty. The samples still need to represent the mix of surfaces in the pixel. A survey of smooth ice alone should not be used to judge a pixel dominated by ridges.

Timing matters just as much as location. A floe can drift, snow can redistribute, and surface conditions can change between a satellite pass and a field visit. Measurements should be as close together in time as practical, ideally near the overpass, with ice motion accounted for.

Calibration is not validation

Suppose researchers measure radar return, surface roughness, and ice-core salinity together. They use those observations to develop a rule for identifying ice types from radar. That is calibration.

Next, they apply the rule to new observations and compare its predictions with independent field measurements. That is validation. Testing only on the measurements used to develop the rule cannot show how well it works elsewhere.

Designing a snow-depth check

To test a satellite snow-depth product:

• Measure the right quantity. Collect snow-probe transects. Measure snow density where needed for the retrieval, and use direct ice measurements to check the site conditions.
• Cover the variety within the pixel. Include smooth ice, rough ice, ridges, and snowdrifts rather than only the easiest walking route.
• Match the satellite observation. Record positions and times, work near the overpass, and track the moving ice.
• Use enough independent coverage. Several representative transects tell us more about a pixel than many closely clustered measurements of the same patch.
• Keep a genuine test set. Do not use every field measurement to tune the retrieval and then call the agreement independent validation.

The exchange works in both directions. Satellite images help choose floes and transects, locate ridges, and identify leads. Field observations then help interpret and check those images.

Inuit knowledge and community observations

Inuit knowledge of sea ice includes distinctions between ice types, the timing of freeze-up and breakup, local currents, travel conditions, and changes across generations. In Nunavut, Inuit Qaujimajatuqangit describes a broader body of knowledge and ways of knowing; sea-ice knowledge is part of it.

This knowledge answers questions that an area-average pixel cannot. A map may show high ice concentration without showing whether a particular crossing is safe. Someone familiar with local currents and recent conditions may recognize a hazard hidden by that average.

SIKU, a platform developed by and for Indigenous communities, brings local observations together with satellite imagery and other information. It supports community-led monitoring and decisions about travel, rather than treating local knowledge only as measurements for outside researchers.[3] Neither a satellite image nor a single thickness measurement guarantees a safe route.

What remains difficult

The hardest conditions to measure are often the ones where good observations are most needed:

• Snow: depth and density vary across a floe and affect thickness retrievals.
• Thin ice: it may be unsafe to stand on, so measurements need remote instruments, poles, drones, or aircraft.
• Deforming ice: ridges and leads can form quickly, making simultaneous measurements difficult.
• Melt season: ponds, wet snow, and changing surfaces complicate both access and satellite interpretation.

A useful way to check your understanding is to explain one complete connection. For example: snow measurements improve satellite ice-thickness estimates because the freeboard-to-thickness calculation depends on how much snow is loading the ice. Naming the instrument is only the start; the important part is explaining why its measurement is needed.

Sources

[1] https://earth.esa.int/eogateway/news/snow-depth-on-sea-ice-among-new-and-improved-cryosat-data-products [2] https://mosaic-expedition.org/expedition/drift [3] https://www.asc-csa.gc.ca/eng/blog/2025/09/29/siku-traditional-knowledge-meets-satellite-data.asp