NadirEO Catalog
Browse and access Earth Observation solutions
Showing 51 of 51
Quantum Computing for Earth Observation
Puts the few Earth observation problems that quantum hardware is genuinely suited to, acquisition scheduling above all, onto quantum processors, as one step inside the normal pipeline.
Quantum Computing for Earth Observation
Begin with what this is not. Quantum processors are not faster at Earth observation in general, and anything implying otherwise is selling something. What they may be better at is a narrow class of mathematically awkward problem that happens to occur in this field. This service is about putting those specific problems, and only those, onto quantum hardware.
The way it works is hybrid, and that matters more than any hardware detail. Classical machines do the bulk of the job (moving, preparing and cleaning large volumes of imagery) and the quantum processor receives only the small, difficult core, handing back a result the classical side assembles. The quantum part is a step inside an existing chain, never a replacement for it, and the hardware is reached as cloud compute rather than bought.
Scheduling and tasking are the first credible tasks for quantum computing: choosing which acquisitions to make and in what order, across conflicting requests and limited capacity, is a combinatorial optimisation problem, and it can be written in exactly the form quantum optimisers accept. It also maps directly onto the acquisition planner already on this roadmap.
Data Encryption
Protects satellite data and the results drawn from it along the whole chain, and lets the customer hold the key to their own material rather than having to trust that nobody else reads it.
Data Encryption
Earth observation data travels further than most data does. It is captured in orbit, sent down to a ground station, archived, processed, combined with other sources, and finally delivered to whoever asked for it. Protecting it therefore means protecting that entire journey, not only the page someone logs into at the end of it. And the moment the subject of an image is sensitive (a border, a port, an industrial site, a piece of critical infrastructure, somebody’s land) both the imagery and the analysis drawn from it become information that has to be handled as such.
What the service provides is straightforward to state. Data is protected while it moves and while it sits still. Results are reachable only by the people entitled to them, and only for as long as they should be. Where the rules require it, processing and storage stay inside an agreed region. And the part that institutional buyers care about most: the customer can hold the key to their own material, so that access to their results becomes something they grant rather than something they trust. Deletion follows the same logic: it can be demonstrated rather than asserted.
There is a wider direction of travel worth being positioned for. Europe is building a quantum-secure communication network spanning the member states, combining terrestrial links with satellites that distribute encryption keys from orbit, the first of which is due to fly this year. The relevant point for a platform is not the physics but the expectation it creates: institutional users are being steered toward key distribution they can trust independently of whoever runs their software. A platform that already keeps key custody separate from the service can connect to that when the time comes. One with keys baked into the application cannot, without being rebuilt.
High-Performance Computing
Runs the platform’s services over a country or a decade of archive, scaling out across many machines, onto GPUs where they are needed, or into a dedicated environment brought up on request.
High-Performance Computing
Every service works over an area of interest. When that area becomes a country, or the period becomes twenty years, the same job is suddenly thousands of scenes, and whether the answer takes two hours or three weeks depends entirely on how much can run at once. The platform is well placed for this: its workers already take jobs from queues, so adding machines raises throughput without changing any service. What is missing is the layer above: splitting one large request into pieces, tracking them as a single job, putting the results back together, and absorbing the failure of a few without losing the run. Some steps are neural and need GPUs; keeping those on accelerated machines while the download-heavy steps stay on cheap ones is what makes a large run affordable rather than merely possible.
Some work cannot share a machine with anyone else’s. For that there are dedicated environments, requested rather than permanently held: an isolated set of machines with its own storage, sized for the job, brought up when the customer asks and taken down when they have finished, billed only for the window it existed. Capacity inside one is reserved rather than best-effort, which is what work with a deadline needs: an emergency response has to run now, not when the shared queue clears. It is also the answer for sensitive work: nobody else scheduled into it, an agreed region, the customer’s own keys. And because the environment is fixed rather than drifting with the platform, the same run repeated a year later runs under the same conditions.
Report Builder
Turns the output of a job into a finished document (maps, figures, statistics, method and sources) under the client’s own cover and branding.
Report Builder
A completed job produces files, e.g.: a GeoTIFF, a shapefile, a link to a map. Sometimes, what the person who commissioned it needs is a document, something to attach to a claim, file with a regulator, hand to a committee or send to a client. Today that conversion happens by hand in a word processor, and it is where a good deal of the value of the analysis quietly disappears into somebody’s afternoon.
From a finished job the builder assembles that document: the result mapped at a sensible scale with a legend and a scale bar, the summary statistics that matter for that particular service, before-and-after imagery where the analysis is a comparison, and the methodological record: which data, which dates, which parameters, which version of the service. That last part is what makes a document defensible rather than merely presentable, and it is the part a person writing under time pressure leaves out.
Each service carries a default template that knows what matters for it: a flood report leads with extent and affected assets, a deforestation report leads with the plot verdict and the date of loss. An organisation can hold its own templates on top: its cover, its typography, its mandatory sections, so that what arrives is already in the form its recipients recognise and its reviewers expect.
Acquisition Planner
Shows when the next usable satellite pass over your area will happen, across every mission at once, and lets you queue a service to run the moment it arrives.
Acquisition Planner
Everyone working with satellite data plans around the next acquisition. An agronomist deciding when advice can be given, an emergency coordinator deciding whether to wait for radar, an analyst committing to a delivery date, all of them need the same fact, and all of them currently assemble it by hand from mission calendars and orbit predictions, one area at a time.
The planner shows a calendar of forthcoming passes over the user’s area for every mission the platform can read, with the orbit direction, the local time of acquisition, how much of the area the swath actually covers, and, for optical missions, the cloud forecast for that hour, so an entry says “probably usable” rather than merely “a pass exists”, which is the distinction that decides whether anyone plans around it.
It is not only a view. Any service on the platform can be attached to a future acquisition: run flood mapping on the next radar pass; run the drought profile on the next optical acquisition below a cloud threshold. The job waits, watches and fires on its own when the condition is met, which turns “check back in a few days” into something the platform does on the user’s behalf.
Oil Spill Detection
Finds oil on the sea surface as a dark patch on radar, by day or night and through cloud, and ties it to the vessel most likely to have left it.
Oil Spill Detection
Oil damps the small capillary waves that make the sea surface bright to radar, so a slick reads as a dark patch against a rough sea. That single physical fact is what makes the service possible at all hours and in any weather: an area of water can be checked on a schedule, regardless of daylight, cloud, or whether a patrol happened to be nearby. Which matters, because a deliberate discharge at night and far offshore is precisely the event that otherwise goes unseen.
Each pass returns the slicks found, with their extent, orientation and estimated area, and a confidence rather than a binary claim. That confidence does real work: plenty of things darken a radar image (low wind, natural biogenic films, algal blooms, rain cells, the wake of a ship) and separating a likely slick from a look-alike depends on its shape, the sharpness of its edges, the surrounding context, and above all the wind conditions at the moment of acquisition. A detection that cannot say how sure it is wastes the responder’s time.
A slick on its own is an observation. What turns it into something actionable is knowing who was there, and the platform already holds the other half: its vessel detection and fusion capabilities answer which ships were in that water at that time, which of them were reporting their position and which were not, and whose track is consistent with the slick’s orientation and apparent age. That is the step between noticing pollution and being able to do something about it.
The benefits follow from those three things. Detection close to the event keeps the response small, because oil spreads and weathers quickly and every hour makes recovery harder and more expensive. Drift projection tells responders where the slick will be when they arrive rather than where it was when it was seen, which is what makes a dispatch decision worth making. Attribution supports recovery of clean-up costs, since the polluter-pays principle only operates once the polluter is identified. Finds oil on the sea surface as a dark patch on radar, by day or night and through cloud, and ties it to the vessel most likely to have left it.
Methane & Air Quality Monitoring
Maps nitrogen dioxide, sulphur dioxide, carbon monoxide and methane across a region, and flags the large methane releases that now carry a reporting obligation.
Methane & Air Quality Monitoring
One instrument family serves two quite different audiences. For environmental and city authorities it is air quality: where the pollution is, how it moves with the season, which corridors and industrial areas stand out. For the energy sector it is methane, and there the status of the question has changed. The EU methane regulation makes measurement, monitoring, reporting and verification mandatory for oil, gas and coal operators and for importers, and provides for a monitoring tool built on satellite data together with a rapid reaction mechanism that notifies states of super-emitting events.
It is worth being exact about what satellites do here. They map trace gases daily and globally at coarse spatial resolution: excellent for regional patterns, plumes and large point sources, useless for a single leaking valve. Super-emitter events are exactly the class they are good at: large, often short-lived, and otherwise invisible until somebody chooses to report them.
The service provides a regional time series with anomaly detection against a seasonal baseline, so that an unusual week reads as unusual rather than as weather; plume delineation with back-tracing through the wind field toward a likely origin; attribution to nearby assets where any exist; and a retained evidence record for each event (imagery, date, plume extent, wind conditions). For the air-quality audience the same machinery produces pollutant maps, seasonal patterns and hotspot rankings over a city or a region.
CAP Area Monitoring
Checks every agricultural parcel in a region against what the farmer declared, continuously through the season, and sorts them into compliant, non-compliant and needs-a-look.
CAP Area Monitoring
Under the reformed Common Agricultural Policy, paying agencies verify farmers’ claims with satellite evidence instead of by sampling field visits: every eligible parcel, every season, monitored continuously from Copernicus data. That moved an entire national administrative process onto Earth observation by regulation rather than by choice, which is what makes it the largest institutional market for this kind of work in Europe.
Given the declared parcels, the service builds a profile of each one through the season from radar and optical imagery, determines the crop group actually grown, and detects whether the required activity took place: mowing, harvest, ploughing, a catch crop established. Radar carries most of the timing questions, because a mowing event is a drop in backscatter that does not wait for a clear sky, and a deadline that falls in a cloudy fortnight is exactly when an optical-only system goes quiet.
Each parcel comes back green where the evidence is consistent with the declaration, red where it contradicts it, and yellow where automated evidence cannot settle it and a human has to look. The yellow rate is the number that decides whether the system saves an agency money: too aggressive and it defends wrong decisions, too cautious and everything queues for an inspector. Confidence per parcel and per marker, with the supporting evidence retained, is what allows that threshold to be tuned deliberately instead of discovered after the season.
Service Builder
Lets someone chain existing NadirEO services into a new one, name it, and publish it to the catalogue, without writing any code.
Service Builder
The platform already runs chains. That mechanism is entirely general: any sequence of platform services can be declared the same way. What is missing is a way for anyone outside the platform team to declare one.
The builder presents the available services as steps, each showing what it needs and what it produces, and lets the author connect them so that the output of one becomes the input or a parameter of the next. Because the orchestrator already validates these definitions, a chain that could not work cannot be drawn in the first place. Each parameter is either fixed by the author or left open to be supplied at launch, which is precisely the difference between a recipe that runs once and a service other people can use.
A finished chain is given a name, a description and an input form, and from that point it behaves like any other service: it appears in the catalogue, it runs through the orchestrator, it is subject to the same authorisation, and its results land in the same place. The author decides whether it stays private, is shared inside their organisation, or is offered to everyone.
The commercial consequence is worth stating plainly, because it is larger than the feature: this changes who is able to create supply. Today a new service means platform work. With a builder, a domain expert who knows their use case is “super-resolve, then detect burnt area, then publish” can assemble it, name it and offer it themselves. It also turns every existing service into a component, which makes the catalogue worth more than the sum of its entries.
Nadira Service Composition
Turns a described need into a running job: Nadira picks the right service, assembles its parameters with you, asks you to confirm, and submits it, then comes back with the result.
Nadira Service Composition
Nadira already answers and recommends. Someone describes a need, she works out whether it is a request for a service, for data or for documentation, and returns ranked service cards or a cited answer. What she cannot yet do is the step the user actually wanted. Between “Flood Mapping is the service for this” and a finished flood map sit a login, a form, an area drawn on a map and a set of parameters and that gap is where most first-time users stop.
The pieces needed to close it already exist, separately. Nadira supplies the part that is missing: the service catalogue and the intent router that work out which flow is the right one in the first place.
The interaction gathers what it can from the conversation, asks only for what is genuinely absent, and then shows the whole request back (service, area, dates, parameters, and the cost or credits it will consume) for explicit confirmation before anything is submitted. Nothing runs implicitly. From there it reports progress and returns the result, or explains in plain language why it failed. These guardrails are not decoration: an assistant that can spend a customer’s compute has to be wrong in ways that are visible and cheap, which means confirm-before-submit, no silent retries, and a record of everything launched on whose behalf.
Nadira Virtual Assistant
NadirEO’s assistant: describe what you need and it points you to the service or the dataset that does it, or answers your question about the platform with its sources cited.
Nadira Virtual Assistant
People arriving at NadirEO could not know which service they need. They know the outcome they want. Nadira takes that outcome in plain language (“I have to map a flood in a river basin”, “I need temperature data for 2050”) and works out whether it is a request for a service, a request for data, or a question about the platform, then answers accordingly: service cards ranked by relevance, matching datasets, or a written answer with citations.
Vehicle Detection & Counting
Counts vehicles at a site and tracks how that number moves over time, turning parking and traffic into a measurable indicator of activity.
Vehicle Detection & Counting
The number of vehicles at a place is one of the most direct observable proxies for what is happening there: cars in a retail park, lorries at a distribution centre or a border crossing, plant on a construction site, vehicles staged at a port. Counted once it is a curiosity. Counted repeatedly it becomes a time series that says whether a site is busier than it was last quarter, with nobody on the ground.
The service works over site polygons the customer defines or the platform derives: car parks, yards, staging areas, road segments. Within them it locates vehicle-like objects and, where they resolve individually, counts them and separates light from heavy. Where they do not resolve individually it measures the aggregate instead (how much of the mapped area is occupied, and at what density) which keeps the indicator continuous rather than leaving a gap whenever the available imagery is coarser.
Building Footprint Extraction
Extracts individual building outlines, with height estimates, over any area, from whatever satellite imagery is available for it.
Building Footprint Extraction
Building footprints are the base layer of exposure modelling, population estimation, energy demand and cadastral work. In much of the world they either do not exist, are years out of date, or stop at the edge of whatever area someone once had mapped. This service produces them on demand, for the area and the date the customer asks for.
It works on objects rather than pixels: segment the image into coherent regions first, then classify each one as building or not. Where the imagery is coarse relative to the buildings in it, the platform’s super-resolution step sharpens it before segmentation, which is what makes footprint extraction workable on medium-resolution sources rather than only on the finest imagery, and is a route the published literature covers up to national scale.
Imagery resolution sets how much detail comes back, not whether the service works. Finer imagery separates adjoining structures and picks up small outbuildings, and turns height from a modelled estimate into a measured one; coarser imagery still yields the built form and the count. Every footprint carries a confidence value, so thin or ambiguous areas are visible as such rather than silently under-mapped, and because extraction is a job like any other it can be repeated on whatever cycle the customer needs.
Vessel Route Intelligence
Reconstructs where a vessel went while it was not reporting, and predicts where it will be, including where to point the next satellite pass.
Vessel Route Intelligence
Two questions, one model. Backwards: a vessel stopped reporting for eleven hours, where could it plausibly have been, and did it pass through anywhere that matters? Forwards: given its current course, history and the routes vessels like it take, where will it be in six hours, and what is the uncertainty around that?
Reconstruction constrains the gap with physics and precedent (the vessel’s speed and turning limits, the bathymetry it can cross, and the corridors that traffic of its type actually uses) producing not a line but a reachable region with probability across it, which is the only honest answer when the data is absent. Prediction runs the same machinery forward, learned from historical traffic, giving destination and arrival estimates with an uncertainty ellipse.
The operationally interesting consequence is acquisition cueing. Satellite passes are scarce and oceans are large; a predicted position with an uncertainty ellipse tells a surveillance operator where to look on the next pass. That closes the loop with vessel detection and turns a passive monitoring chain into a directed one, which is the difference between watching the sea and following a ship.
Vessel Anomaly Detection
Spots the behaviours that precede an offence (going dark, faking a position, loitering, meeting another ship at sea) and ranks them for attention.
Vessel Anomaly Detection
Illegal activity at sea has a small vocabulary of behaviours. A transponder switches off just before entering a closed fishing area and back on after leaving it. A reported position drifts somewhere the vessel physically cannot be. Two ships slow, converge and sit together for hours in open water. A vessel loiters at the edge of a jurisdiction. None of these is proof of anything, and all of them are worth a look.
The service watches for that vocabulary across the fused traffic picture and produces prioritised, explained alerts: what pattern triggered, where, when, and what corroborating imagery exists. Explanation matters more than detection here, so every alert carries the evidence that produced it.
Vessel Tracking & Multi-Source Data Fusion
Fuses radar, optical, radio-frequency and night-light detections with self-reported positions (AIS), vessel registries and the marine environment into one identified traffic picture and flags everything that does not add up.
Vessel Tracking & Multi-Source Data Fusion
Detection tells you a ship is there. Fusion tells you who it is and, far more usefully, which ships are there without saying so. No single source manages that on its own, because every source has a blind spot the others cover. The work of this service is to bring all of them onto one timeline, one geometry and one identity, and then to report what is left over.
Four sensing layers go in. Self-reported positions come from AIS, terrestrial and satellite, and where the customer is an authority that holds them, from long-range identification and tracking reports and from fisheries vessel monitoring systems. Imaging detections come from the platform’s own radar and optical services, which between them cover night, weather and the small non-metallic hulls radar cannot see. Radio-frequency geolocation finds vessels by their emissions (navigation radar, VHF, satellite telephone) which works precisely when a transponder has been switched off. Night-time light detection finds light-luring fishing fleets that neither radar nor daylight optical reliably catch.
Two more layers turn positions into meaning. Registries resolve a detection into a named vessel: the international ship registry, national and EU fleet registers, the authorised-vessel and illegal-fishing lists maintained by regional fisheries bodies, sanctions designations, ownership and flag history, and port state control inspection records. Ancillary marine data supplies the context that makes a position interpretable, EMODnet for bathymetry, vessel density, wind farms, cables, pipelines and extraction sites; exclusive economic zones, territorial waters, protected areas, traffic separation schemes and port polygons for jurisdiction; and met-ocean fields from the Copernicus marine service for wind, waves, currents and sea ice. Two of these do real analytical work rather than decoration: bathymetry rules out positions a hull of a given draft physically cannot occupy, and sea state governs how much the radar could have missed, which is what turns a completeness statement from a disclaimer into a number.
The output is a traffic picture where every position carries its provenance, which source saw it, when, and with what confidence, and every detection is classified as matched, unmatched or ambiguous. The unmatched ones are the operational product.
Optical Vessel Detection
Sees the boats radar misses (wooden, fibreglass and inflatable hulls) and at high resolution says what type of vessel each one is.
Optical Vessel Detection
Radar detects conductivity. Optical detects contrast. That difference decides which vessels each sensor can find, and the gap matters more than it sounds: wooden fishing skiffs, fibreglass RIBs and inflatables (the hulls that dominate small-scale illegal fishing, coastal smuggling and irregular movement at sea) are weak radar targets and frequently invisible to it. To an optical sensor they are ordinary targets.
The service detects vessels and their wakes in the visible and near-infrared bands, rejecting the two things that fool a naive detector (sun glint and whitecaps) and reporting honestly on cloud, because a clouded pass is not an empty sea. One property of Sentinel-2 is worth exploiting explicitly: its spectral bands are not acquired simultaneously but with offsets of up to about 2.6 seconds, so a moving vessel sits at slightly different positions in different bands. That displacement yields speed and heading from a single image, a measurement radar needs a visible wake to make.
Resolution sets what the answer can be. At Sentinel-2’s 10 metres, free and every five days, the service finds larger vessels and wakes and measures their motion. At sub-metre, commercial imagery resolves hull shape, superstructure and deck arrangement, which is what allows a vessel to be classified (fishing, cargo, pleasure, patrol) rather than merely counted and measured.
Radar Vessel Detection
Finds metal-hulled ships across an area of ocean by day or night and through any weather, with size, heading and confidence for each.
Radar Vessel Detection
A metal hull on water is close to an ideal radar target: bright, compact, against a dark and largely uniform background. That physics is why radar remains the foundation of maritime surveillance from space, and why detection works equally at three in the morning, under a storm, and through the Arctic winter. It is the only sensor that gives a genuinely persistent picture of a sea area.
The service screens SAR passes over an area, detects targets against the local sea clutter rather than against a fixed threshold, masks land, platforms and wind farms, and estimates each target’s length, width and heading. Where a wake is resolved it adds a speed and course estimate. Output is a georeferenced detection list per pass, with confidence per detection and a completeness statement covering sea state, because in a rough sea small vessels are missed, and a user needs to be told that rather than shown an empty ocean.
Asset Climate Resilience
Tells an asset owner the year each site's design basis stops being enough, and ranks the portfolio by how soon.
Asset Climate Resilience
Every solar plant, wind farm and substation was engineered against a design event (a wind gust, a temperature, a hail size) drawn from the climate record available when it was built. Climate change moves that record. At some point the design event stops being the once-in-fifty-years case and becomes ordinary. This service identifies when: the point of no return for each site in a portfolio.
The engine already exists and works. It fits a non-stationary extreme value distribution to the full record of annual maxima, with the distribution’s location shifting through time, producing a smooth return-level trajectory that is far more stable than re-fitting rolling windows. Comparing the future return level against the plant’s design value gives both the year of exceedance and the change in design load, which for wind scales with the square of velocity under the standard engineering codes.
The natural extension is to the hazard that actually destroys solar plants. Hail causes around three-quarters of utility-scale solar losses by value from a small minority of claims, and insurers are repricing accordingly. Adding hail exposure and heat derating alongside wind turns a single-hazard study into the portfolio climate risk assessment that lenders and insurers are now asking owners to produce.
Renewable Siting & Planning
Finds and ranks the parcels where a renewable project is both physically viable and legally permissible, and shows why each one qualified.
Renewable Siting & Planning
Siting is an exclusion problem before it is an optimisation problem. Most land is unavailable (protected habitat, too steep, too close to housing, inside an airport zone, on floodplain, wrong land cover) and the work is in applying every constraint consistently across a territory rather than arguing about weights. This service does the exclusion first, transparently, then ranks what survives.
Constraint layers are configurable because the rules differ by jurisdiction and by technology: protected-area networks, slope thresholds, settlement buffers, aviation and radar zones, flood and landslide exposure, land cover classes, grid and road distance. Weighted multi-criteria scoring then ranks the remaining candidates, and every candidate carries an audit trail of which constraints it passed and on what data, which is what makes the result defensible in a planning process rather than merely persuasive.
Expected Energy Yield Map
Turns resource into electricity: a map of how many megawatt-hours a specific plant configuration would actually generate, location by location.
Expected Energy Yield Map
Irradiance is not energy. Between the two sit the choices that decide a project’s economics: fixed tilt or single-axis tracking, module technology, bifacial gain, inverter loading, temperature derating, soiling, snow, and the horizon that the surrounding terrain imposes. This service applies those choices and outputs the quantity that a business case is built on: expected generation.
For solar it runs a full simulation chain per pixel, producing specific yield in kilowatt-hours per kilowatt-peak and absolute generation for a stated capacity, with monthly and hourly profiles. For wind it applies a chosen turbine’s power curve to the modelled wind distribution at hub height, with wake and availability assumptions stated rather than buried. Both carry the same terrain-derived horizon shading, which is the factor most desk studies omit and most disappointing sites turn out to have.
The output is a heatmap, which matters more than it sounds: seeing yield vary across a candidate parcel is what tells a developer which corner of it to build on, and what a single site-level number never shows.
Renewable Energy Potential Index
Scores any area for solar and wind resource quality on one comparable index, built from two decades of satellite and reanalysis records.
Renewable Energy Potential Index
Before anyone models a plant, they need to know whether a place is worth modelling. This service answers that: a single normalised index per location, derived from long-term solar irradiance and wind resource records, so that candidate areas across a region, or across countries, can be ranked on the same scale in one pass.
The underlying quantities are: global horizontal and direct normal irradiance for solar, wind speed and power density at the relevant hub heights for wind, both from multi-decade satellite-derived and reanalysis datasets, terrain-corrected. Because the records are long, the index carries not only a central estimate but the variability around it.
What turns resource data into an index is combining it with the factors that make good resource usable: terrain slope, distance to grid, distance to road.
Urban Expansion Monitoring
Measures how far and how fast a settlement has grown, and how much of that growth went into hazard-exposed land.
Urban Expansion Monitoring
The service maps built-up extent at successive epochs and turns the difference into the numbers a planner or a development bank actually asks for: hectares added, rate of expansion, direction of growth, and the share of new settlement that landed on floodplain, steep slope or protected land.
It combines radar and optical: new construction changes the structure of a surface before it changes its colour, and radar sees that structural change through cloud and at night.
Landslide Motion Monitoring
Measures millimetre-per-year slope movement from radar, and flags the slopes that are accelerating.
Landslide Motion Monitoring
Most slopes that fail were moving first, often for years, at rates of millimetres per year that nothing on the ground detects. Radar interferometry measures exactly that, from orbit, over whole regions, without instrumenting anything. The service builds a displacement time series for every coherent point on a slope, produces a velocity map, and flags where velocity is increasing, because acceleration rather than velocity is what precedes failure.
Combining ascending and descending passes separates vertical from east-west movement, which distinguishes a slope creeping downhill from ground subsiding beneath it. Layering deformation over terrain, land cover and rainfall history yields a susceptibility model that is dynamic rather than static: not just where a landslide could occur, but where the ground is currently telling you it might.
Wildfire Probability
Maps where the landscape is primed to burn, at field scale, by combining live fuel condition from satellite with fire weather and terrain.
Wildfire Probability
Europe’s operational fire danger forecast runs from meteorological models at around 8 to 10 kilometres. That is the right scale for a national warning and the wrong scale for deciding where to pre-position a crew, close a road or clear a firebreak. This service downscales the picture by adding what satellites see and weather models cannot: the actual state of the fuel.
Live fuel moisture is estimated from Sentinel-2 vegetation and water indices together with thermal data, fuel type and load come from land cover, and terrain contributes slope and aspect, which govern how fire spreads once started. These combine with the meteorological fire weather index and with ignition-proximity factors (roads, settlements, power lines) into a probability surface at tens of metres rather than kilometres.
Two modes serve two decisions: a seasonal susceptibility map for planning fuel management and resource placement, and a short-horizon danger map that updates as fuel dries and weather turns. The honest framing is important, this ranks relative likelihood across a landscape very well, and it will not tell anyone that a specific hectare will burn on a specific day.
Land Cover Change
Reports what turned into what between two dates (e.g.: forest to cropland, cropland to built-up) with the area of every transition.
Land Cover Change
Knowing that something changed is rarely the question. The question is what it became: how many hectares of forest became pasture, how much farmland was built on, where wetland was drained. This service produces that answer as a transition matrix — every from-class to to-class pair, with its area and its location.
It works by classifying two epochs consistently and then comparing them, rather than by differencing images directly, which is what prevents seasonal appearance from masquerading as land use change.
Because the same method runs over any pair of dates, it serves both directions of a typical request: the retrospective study that establishes what has happened over a decade, and the recurring report that says what changed since the last one.
Land Cover Fusion
Produces a current 10-metre land cover map for any area by combining radar and optical satellites, in the class scheme the customer actually uses.
Land Cover Fusion
Global land cover products exist and are free, but they are fixed: fixed classes, fixed epochs, fixed accuracy, and never quite the classes a given customer’s regulation or business asks for. This service produces land cover to order, for the area, the date window and the class scheme requested.
Its accuracy comes from fusion. Optical imagery carries the spectral information that distinguishes crops from grass from trees; radar carries structure and, crucially, is unaffected by cloud, which is how a map gets made over a region that is overcast for half the year. Combining Sentinel-1 and Sentinel-2 across a season also exploits phenology: a winter cereal and a summer crop are identical in one image and unmistakable across twelve.
Where the customer has their own nomenclature (e.g.: a national scheme, a Corine-compatible legend, a bespoke set of classes for a permit) the service trains against their reference polygons rather than forcing them onto someone else’s classes. Every output carries a per-pixel confidence layer, because a land cover map without one invites more trust than it has earned.
Corridor Vegetation Encroachment
Watches vegetation growing into power lines, railways, roads and pipelines, and ranks which stretches need cutting first.
Corridor Vegetation Encroachment
Vegetation growing into an infrastructure corridor is a slow, predictable, expensive problem. On power lines it causes outages and starts fires; on railways it obscures signals and drops branches on the line; on pipelines it hides the right of way and roots into it. Operators manage it by cutting on a cycle, which means cutting where nothing needed cutting and missing where it did.
The service replaces the cycle with evidence. Given the geometry of the asset, it builds a corridor buffer, tracks vegetation vigour and height proxies inside it through the season, computes a growth rate per segment, and ranks segments by how close they are to breaching clearance.
The product is triage that tells a crew where to go, not a replacement for the inspection they do when they get there.
Deforestation Alerts
Proves, plot by plot, that a commodity was not grown on land deforested after 2020 — the evidence the EU Deforestation Regulation requires.
Deforestation Alerts
Given a set of plot boundaries, the service reconstructs forest cover from the archive, then screens the full period since for loss, using radar alerts that see through the tropical cloud that defeats optical monitoring, confirmed against optical imagery when the sky clears. Each plot comes back with a verdict, the date of any detected loss, the area involved, and the imagery that supports it.
The output is built with a per-plot compliance record with a traceable evidence pack, retained for the five years the regulation demands. The same machinery serves voluntary commitments and lender reporting, but the regulation is what puts a deadline on it.
Agricultural Drought Watch
Drought is not visible as an absolute value, a field can look green and still be in trouble. What matters…
Agricultural Drought Watch
Sign in or create an account to access this product.
Seamless Mosaicking
Builds one seamless, cloud-free image over an area of any size, from however many scenes, dates and orbits it takes to cover it.
Seamless Mosaicking
Real areas of interest rarely fit inside one satellite scene, and the scenes that do cover them are rarely cloud-free on the same day. The result, without this service, is a patchwork: visible tile edges, mismatched colour between acquisition dates, and holes where the cloud was. Mosaicking is what turns that into a single image someone can actually put in a report.
The service selects, for every output pixel, the best available observation across the requested period, using cloud and shadow masks, view angle and recency, then harmonises radiometry between contributing scenes and feathers the seams so no tile boundary survives into the product. Compositing rules are chosen per use: most recent cloud-free for currency, median for stability, minimum-cloud for clarity.
Optical Co-registration
Aligns optical imagery from any sensor, date or resolution onto one common grid, so that change detection measures change and not misalignment.
Optical Co-registration
Every optical image arrives georeferenced and almost none arrive perfectly co-registered. Residual orbit and attitude errors, terrain parallax on off-nadir acquisitions, different elevation models used in orthorectification, and successive processing baselines all leave a shift behind, typically a fraction of a pixel, often several. For a mosaic that is cosmetic. For change detection it is fatal: a one-pixel shift along a field boundary or a roofline produces a bright edge of false change that no threshold can separate from the real thing.
This service removes that shift, whatever produced the image. It aligns each scene to a chosen reference, an external reference dataset where one exists for that sensor, a master scene from the stack, a stack median, or the customer’s own orthophoto or cadastral base, by matching stable features and estimating the offset to sub-pixel precision. Where a single global shift is not enough, because terrain parallax or a strongly off-nadir geometry distorts the scene locally, it fits a local warp field instead. Matching works across sensors and across resolutions, so a metre-class commercial acquisition, a Sentinel-2 scene, a drone orthophoto and a decades-old scanned archive frame can be brought onto the same grid as the open medium-resolution archive.
The output carries a quality report stating the residual shift actually achieved, per scene, in metres and in pixels, so a downstream analyst knows whether a detected change is larger than the registration error. That number is the difference between a result and a claim. It is also what makes a genuinely multi-sensor time series possible: one coherent record spanning whatever imagery the customer can get, rather than several incompatible ones.
SAR Coherence Stack
Aligns a time series of radar acquisitions to a fraction of a pixel, the precondition for measuring anything that moves.
SAR Coherence Stack
Interferometric coherence answers a question no optical index can: did the physical arrangement of this surface stay the same between two dates? A parked car park, a roof, a road, these stay coherent for years. A ploughed field, a flooded plain, a demolished building, a new excavation, these lose coherence instantly. Coherence is therefore a change detector that is indifferent to cloud, to darkness, and to seasonal greening that fools NDVI.
The service computes coherence for every pair in a co-registered stack, then derives the products that make it usable: a mean coherence map showing what is permanently stable, a coherence change map flagging where stability broke and when, and per-pixel time series. Because the measurement is relative, it works without any training data and without a labelled baseline.
New construction, demolition, flood under cloud, deforestation confirmation, urban stability and crop harvest tracking all read the same underlying signal.
SAR Terrain Geocoding
Turns a raw Sentinel-1 acquisition into a calibrated, terrain-corrected radar image on a map grid, ready to analyse or stack.
SAR Terrain Geocoding
Raw radar is not a map. It is recorded in the geometry of the satellite’s own line of sight, distorted by terrain, uncalibrated and speckled. Everything a user actually wants from SAR (e.g.: flood extent, ground motion, vessel positions, change) starts by undoing that. This service does exactly that and nothing more, and it does it the same way every time.
The chain applies precise orbit files, removes thermal noise and border artefacts, calibrates to a physical backscatter coefficient, then performs Range-Doppler terrain correction against specific DEM so each pixel lands where it belongs on the ground. Optional speckle filtering and decibel conversion round it off, and the result is written as a cloud-optimised GeoTIFF clipped to the area of interest.
Geosphere: Ground Motion
Visualises millimetre-scale ground motion from satellite interferometry, with the displacement time series of every single measurement point.
Geosphere: Ground Motion
Geosphere is the WebGIS client for PS-InSAR ground deformation data based on the European Ground Motion Service. Draw the area, pick level, orbit and motion component, and the result is a map of measurement points coloured by displacement velocity: subsidence, uplift, slope instability, settlement near major infrastructure.
Change Detection
Finds real changes on the ground between two dates, separating them from the false changes caused by shadows and seasonal vegetation.
Change Detection
The service compares two super-resolved Sentinel-2 images of the same area acquired on different dates and returns polygons of what actually changed: new construction, demolition, earthworks, land-use conversion.
The initial change map comes from principal component analysis followed by K-means clustering on the difference image — an unsupervised approach that needs no training data. The real value, though, is in what comes next: the two great producers of false change — shadows, which move with the solar angle between acquisitions, and vegetation, which changes with the season — are masked explicitly. Shadows are detected with multiple thresholds and then projected according to each scene’s actual solar angle, so a shadow shifting position is never mistaken for a change on the ground.
Burnt Area Detection
Delineates the area burnt by a wildfire from Sentinel-2 imagery before and after the fire, cross-checked at native and super-resolved scale.
Burnt Area Detection
The application produces the perimeter of a burnt area after a wildfire, from a pair of Sentinel-2 images acquired before and after the event. The result is a vector layer ready for damage assessment and restoration planning.
The indicator is dNBR, the difference in the normalised burn ratio derived from the near-infrared and short-wave infrared bands — the most reliable spectral signature of charred vegetation. What sets it apart is the double pass: dNBR is computed both on the native Sentinel-2 pair and on the super-resolved pair, and the final perimeter is the intersection of the two. The two resolutions validate each other, cutting the false positives a single threshold would produce.
The perimeter is then cleaned of isolated fragments and smoothed before export.
Sustax – Climate Intelligence
Sustax (by Geoskop) delivers highly accurate, validated, decision-ready global climate data — historical reanalysis and forward-looking scenario projections — as standardised, ready-to-use datasets for climate-risk and impact analysis.
Sustax – Climate Intelligence
Sustax, developed by our partner Geoskop, is a climate-data platform that turns raw climate information into accessible, standardised, decision-ready datasets. It combines CMIP6 climate projections with ERA5 reanalysis, so you get both historical observations and future scenarios across the IPCC Shared Socio-Economic Pathways (SSP1-1.9 … SSP5-8.5).
Data is available as daily variables and monthly climate indices across five domains — temperature, precipitation, wind gust, relative humidity, and solar radiation — on a global grid. ERA5 covers 1979–2022 and the CMIP6 projections run to 2080. Every dataset is bias-corrected and validated against ERA5, with accuracy metrics and model-spread information provided for transparency about uncertainty.
On NadirEO you request any Sustax dataset for your location through the platform API; the service returns a time-series export you can integrate directly into your risk and impact workflows.
Data Access
NadirEO unified access to over 200 open Earth data sources — satellite, climate, elevation, land cover, and more, with on-the-fly search, clipping, and delivery.
Data Access
NadirEO Data Access is the platform’s unified gateway to Earth data, providing resilient, on-demand access to a broad catalogue of more than 200 open sources — spanning far beyond satellite imagery. Alongside optical and radar missions such as Sentinel-1, Sentinel-2, and Landsat, the service reaches climate and weather reanalysis, elevation and terrain models, land cover and land-use datasets, vegetation and biomass products, population and built-environment layers, and a wide range of environmental and geophysical datasets.
Built on cloud-native geospatial standards, the service discovers data through STAC catalogues and reads directly from public cloud storage, returning only the area and variables requested rather than depending on the rate-limited APIs of individual providers. Users simply select the dataset, area of interest, time range, bands or variables, and output format; the service transparently handles spatial clipping, multi-scene mosaicking, format conversion, and reliable delivery at scale, combining unrestricted free sources with a managed queue for rate-limited archives.
Fully integrated with the NadirEO ecosystem, Data Access acts as the data foundation for the platform’s application modules and services — from super-resolution and flood delineation to hazard assessment — feeding analysis-ready data directly into automated Earth Observation workflows, or delivering it to your own tools and pipelines through a developer-friendly API. By removing the complexity of multi-source data retrieval, it lets teams focus on analysis and decision-making rather than data logistics.
Flood Mapping
An end-to-end service that maps the extent of a flood by comparing Sentinel-1 radar images taken before and after the event, and publishes the result as a browsable map.
Flood Mapping
An end-to-end service that produces a flood map on demand, starting only from an area of interest and the date of a flood event. The service automatically requests the required Sentinel-1 radar imagery through the platform’s Data Access layer, runs the flood delineation to extract the extent of the inundated areas from before- and after-event radar acquisitions, and publishes the resulting flood map directly to GeoNode, where it can be explored, shared, and combined with other layers. By orchestrating data retrieval, processing, and publication into a single automated workflow, it removes the need for manual data search and processing, delivering a ready-to-use flood-extent map shortly after the event. Because it relies on radar data, mapping is possible even through cloud cover, making the service well suited to rapid post-event response, damage assessment, and insurance claim verification.
Flood Delineation
Earth Observation application that extract the extent of flood events using radar satellite imagery.
Flood Delineation
Earth Observation application that maps the extent of flood events using radar satellite imagery. By comparing pre- and post-event acquisitions, the service delineates inundated areas and identifies affected assets and land, supporting damage assessment, and insurance claim verification. Radar data enables mapping even under cloud cover. Outputs are delivered as flood-extent layers ready to use.
Burnt Area Mapping
Earth Observation service mapping the extent and severity of wildfire events using satellite imagery. By analysing pre- and post-event acquisitions, the service delineates burned areas, estimates burn severity, and quantifies the affected surface, supporting impact assessment, damage estimation, and post-event recovery planning. Outputs are delivered as burned-area perimeters and severity maps ready to use.
Space Academy
A catalogue of space training services built on 60+ years of operational experience, delivered by global space experts.
Space Academy
“The academy features a catalogue of space training services that is built upon over 60 years of operational experience across the full range of space sector disciplines, delivered by global space experts with decades of experience.
Training spans subjects including Earth observation, satellite navigation, satellite communications, ground segment operations and engineering, satellite operations, space engineering management, and space business-related disciplines. A combination of hands-on and theoretical training can be delivered, and can be targeted at different levels (beginner, intermediate, advanced).
Training programmes can be customised to your needs and are delivered by a team of international experts from world-leading international facilities in Europe (Italy, Germany, the Netherlands, and the UK), or delivered locally subject to the availability of the necessary infrastructure, facilities, and resources. Where training is delivered locally on the client’s premises, the team works with the client to ensure the facility is suitably equipped to deliver the training.”
JupyterHub
Collaborative platform to run Jupyter Notebooks on shared infrastructure for EO data processing and analysis.
JupyterHub
JupyterHub is a collaborative platform that allows multiple users to access and run Jupyter Notebooks through a shared web interface. It is built to support teams working on data science, research, or education by providing isolated environments for each user while running on centralized infrastructure.
In the context of Earth Observation, JupyterHub is especially useful for processing satellite data, running geospatial analyses, and sharing reproducible workflows. It enables scientists, developers, and analysts to work together efficiently, using powerful computing resources and standardized tools, all accessible from a browser.
GeoNode
Open-source geospatial content management system to publish, share, and collaboratively use geospatial data and maps.
GeoNode
GeoNode is an open-source geospatial content management system – a platform for the management and publication of geospatial data. It brings together mature and stable open-source software projects under a consistent and easy-to-use interface, allowing non-specialised users to share data and create interactive maps. Built-in data management tools support the integrated creation of data, metadata, and map visualisations, and each dataset can be shared publicly or restricted to specific users. As a flexible and extensible platform (an OSGeo project), it can be customised and integrated with other applications to meet specific requirements.
ArcGIS Enterprise
Comprehensive Esri GIS platform to manage, analyse, and share spatial data in a secure, scalable infrastructure.
ArcGIS Enterprise
ArcGIS Enterprise is a comprehensive geographic information system (GIS) platform developed by Esri that allows organizations to manage, analyze, and share spatial data within a secure, scalable infrastructure. It is designed for deployment on-premises or in cloud environments and is often used by governments, utilities, transportation agencies, and other organizations that rely heavily on geospatial data. Within the platform, it provides the GIS environment used to build interactive dashboards and analytical products over Earth Observation outputs – for example asset condition indices, hazard dashboards, and transportation-asset management views.
GeoReporting
Collaborative environment to integrate datasets, visualise geospatial information, and generate Earth Observation reports.
GeoReporting
This collaborative solution simplifies and enhances the creation of Earth Observation reports. Teams can seamlessly integrate diverse datasets, visualize geospatial information, and generate insightful reports using intuitive, efficient tools. The platform streamlines the entire workflow, from initial data ingestion to the final report output. By fostering teamwork and providing user-friendly features, this system empowers organizations to produce high-quality Earth Observation reports with greater clarity and speed, ultimately improving understanding and decision-making based on valuable Earth insights.
Airport Monitoring
Satellite-based monitoring of airport infrastructure for safety, efficiency, and regulatory compliance.
Airport Monitoring
Space geospatial intelligence plays a crucial role in enhancing airport safety, efficiency, and sustainability. By leveraging satellite data and advanced analytics, airports can monitor infrastructure conditions, detect anomalies, and ensure regulatory compliance, supporting smarter, more resilient, and data-driven operations.
The service combines multiple satellites and sensors (radar, optical, and multispectral) to monitor airport infrastructure performance together with its surrounding environment. Historical satellite analysis allows past data to be reviewed to identify previously unknown issues, such as differential settlement of structures and runways or unauthorised vegetation and building encroachments. AI combines regulations, best practices, satellite data, asset characteristics, and client constraints to improve the identification of critical issues and prioritise actions and interventions. The approach is applicable and scalable for any type of airport asset operator, optimising costs and increasing return on investment for the monitored areas.
Solar Panel Detection
Automatic detection and mapping of solar panels from satellite imagery using AI.
Solar Panel Detection
Solar Panel Tracker is an Earth Observation service designed to automatically detect and map solar panels using satellite imagery. Leveraging advanced image analysis and AI algorithms, it identifies photovoltaic installations across both urban and rural areas, producing a geolocated inventory of solar assets over the area of interest. The application supports energy infrastructure monitoring, planning, and sustainability assessments – for example mapping the spread of distributed photovoltaic capacity, supporting grid and asset planning, tracking the growth of installations over time, and informing renewable-energy and sustainability reporting. Outputs are delivered as georeferenced layers ready for integration into GIS and reporting workflows or directly on Geonode application to visualise them.
Sentinel-2 Super Resolution – 1m
API-based access to Gamma Earth super-resolution for S2 imagery from 10 m to 1 m spatial resolution.
Sentinel-2 Super Resolution – 1m
An Earth Observation pre-processing module that enhances the spatial resolution of Sentinel-2 imagery, producing 1-metre super-resolved images from standard 10-metre products.The solution is designed to submit satellite imagery directly to Gamma Earth’s processing pipeline, retrieve enhanced outputs programmatically, and integrate results into broader EO workflows or visualization platforms. This API-based approach ensures scalability, automation, and flexibility for Earth Observation applications, particularly in environmental monitoring, disaster response, and land analysis. Thanks to its modular and developer-friendly architecture, the solution can be easily integrated into third-party applications, enabling developers to embed Gamma Earth’s capabilities directly into their own tools, dashboards, or data pipelines.
Sentinel-2 Super Resolution – 5m
Single-image super-resolution that enhances Sentinel-2 imagery from 10 m to 5 m spatial resolution.
Sentinel-2 Super Resolution – 5m
An Earth Observation pre-processing module that enhances the spatial resolution of Sentinel-2 imagery, producing 5-metre super-resolved images from standard 10-metre products. The module reconstructs ten Sentinel-2 bands – B02, B03, B04, B05, B06, B07, B08, B8A, B11 and B12 – at 5-metre resolution from L1C or L2A inputs. By increasing the effective level of detail, it improves the quality of downstream analyses such as object detection, land cover classification, and change detection, and integrates directly into Earth Observation processing workflows on the platform.
