Choosing the wrong source doesn’t mean getting a less sharp image. It means not having the data the problem requires.

Spatial Resolution: Chosen by the Problem, Not the Budget

In satellite monitoring, choosing the wrong source doesn’t produce a less sharp image. It produces the absence of the very data the problem requires — a distinction the resolution metric alone doesn’t make obvious.

What changes with sub-meter resolution

With a 50-centimeter pixel, that’s enough for the difference between canopy within or beyond the limit to become distinguishable. That granularity opens questions moderate resolution can’t answer: is the tree 8 or 12 meters from the conductor? Is that ground movement the start of a creep or agricultural prep? Was pruning actually carried out on the stretch the team reported?

Sub-meter resolution, from this angle, isn’t just a quality upgrade. It’s a shift in the class of problem the data can solve.

Resolution is a function of the event, not the budget

The most common mistake in a monitoring contract is treating resolution as a cost variable. Following a purchasing logic where high resolution costs more, moderate costs less, and the choice follows the available budget. That logic tends to produce systems that monitor without detecting what they need to detect, because the question that should guide the choice is a different one: what’s the smallest event that needs to be detected for the monitoring to have value?

When the answer is: vegetation above 5 meters near the conductor, a 10-meter pixel usually handles it.

When the answer is: construction in its early stages within the right-of-way, no budget closes that problem with a 10-meter pixel. The right architecture, therefore, doesn’t pick one resolution to apply across the whole portfolio. It matches the source to the type of event each segment of the asset needs to detect.

How resolution and revisit complement each other

Spatial resolution and temporal resolution aren’t independent variables, and the balance between them shifts depending on the speed of the event you want to capture. A sensor with a 50-centimeter pixel and monthly revisit detects the event in detail, but late. A sensor with a 10-meter pixel and daily revisit detects fast changes without the detail needed to classify them.

Slow-progressing events, like vegetation growth, tolerate a more spaced-out revisit, and what matters there is resolution capable of distinguishing the event from its surroundings. Fast-progressing events, like the start of construction or a fire ignition, flip that priority: what matters is the interval between images. Between the two extremes sit small, fast events, like material dumped in a right-of-way, which require resolution and revisit simultaneously — and are exactly what single-source contracts leave uncovered most often.

What clouds do to any optical resolution

Resolution and revisit lose relevance when the sensor simply can’t see the ground. In tropical Brazil, persistent cloud cover can make the optical series unusable for two to four consecutive weeks during the rainy season — a limitation no pixel specification solves.

SAR, synthetic aperture radar, solves this blind spot by emitting microwave pulses that penetrate clouds, smoke, and darkness and return with information on ground structure and roughness. In lower-frequency bands, the signal can even penetrate the vegetation canopy and capture what’s underneath it, but pays a price for that: it delivers no color or spectral texture, so classifying vegetation type precisely requires combining it with multispectral optical. No single sensor covers tropical Brazil without leaving a blind window. Continuity of coverage, in this context, is always the product of combining sources, each covering the other’s blind spot.

How NOR architects the combination of sources

NOR doesn’t apply a single source across the entire portfolio. The architecture is configured by asset type, event profile, and regulatory requirement.

For linear assets, like transmission lines and pipelines, the base combines high-revisit multispectral optical, for detecting changes at scale, with SAR for continuity during cloudy periods, and sub-meter imagery comes in on demand to confirm events that moderate resolution detected but can’t classify with the precision needed to open a ticket.

NOR Super Resolution: The Algorithm That Turns Free Imagery Into Sub-Meter-Precision Data

In satellite monitoring, 10 meters is just 1 pixel — sharp enough for a large wildfire but not enough to confirm whether clearing was carried out or whether a tree canopy crossed the regulatory limit. Sub-meter resolution solves this problem, but has historically carried a cost that grows with the size of the portfolio: dedicated tasking, area by area, is too expensive to apply at the scale of a ten-thousand-kilometer grid or an entire mining complex.

This is the kind of choice usually treated as a budget dilemma. In practice, it’s an engineering problem NOR already has a solution for.

The problem super-resolution attacks

The largest-coverage free optical source today operates at a native resolution around 10 meters, updated with frequent revisits and no acquisition cost. It’s abundant data, but too coarse for the events that matter most on a linear or mining asset: the start of construction, localized ground movement, material dumped in a restricted strip. The traditional alternative is replacing that source with tasked sub-meter imagery, which solves the detail problem but brings back the cost-at-scale problem.

NOR resolves this tension by processing the free imagery through NOR Super Resolution, a proprietary AI-based super-resolution algorithm that reconstructs detail below the original source’s native pixel. The result takes free 10-meter optical capture to an effective resolution of 2.5 meters, with no dependence on dedicated tasking or paid sub-meter sensors for every scene.

What this changes in practice

Applied at scale, Super Resolution repositions what’s economically viable to monitor. An entire portfolio can now be scanned at near-sub-meter granularity using the same free, high-revisit source that already covers the whole planet, and finer native-resolution tasked imagery comes in only where confirming a specific event requires detail that even super-resolution can’t reconstruct, like fine texture reading in edge cases.

This combination changes the architecture logic described earlier in this piece: instead of choosing between broad coverage and fine detail, the portfolio now operates with both properties by default, reserving on-demand capture for the fraction of events that genuinely require confirmation at native resolution.

Why this matters for operators of linear or mining assets

For a transmission company, a mining company, or a pipeline operator, the practical effect is that cost stops growing proportionally with the asset’s extent. High-granularity coverage, once reserved for select critical areas because of tasking cost, becomes viable across the entire portfolio, bringing detection of small events, like early-stage illegal construction or localized ground movement, closer to the same continuous-monitoring cadence already applied to larger-scale events.

NOR Super Resolution operates integrated into NOR’s same multi-source architecture, combining high-revisit optical, SAR for continuity during cloudy periods, and now AI-reconstructed resolution as a standard layer, with tasked imagery reserved for cases where confirmation requires the detail only native capture delivers.