Geocartis carries out drone surveys for railway projects across India: corridor topographic survey for doubling, gauge conversion and new line work, station and yard mapping, railway land and encroachment documentation, earthwork quantities, and monthly construction progress capture. Work is at 5 cm ground sample distance or better, tied to DGPS control along the alignment, with an RMSE report against every dataset. We are based in Ahmedabad and mobilise to project sections nationally.
We are a survey and geospatial services company. We do not build drones and we do not sell software.
Where drone survey fits in railway work, and where it does not
Railway survey in India covers a wider range of work than most clients realise, and drone photogrammetry is the right tool for some of it and the wrong tool for the rest. Setting that out first saves everybody time.
The Railway Board has adopted satellite imagery, drone photogrammetry and LiDAR as survey technologies for railway projects, and the Indian Railways Construction Manual now reflects that. Which of the three applies depends on the stage and the corridor.
For long greenfield alignment surveys at DPR stage, the method that has become standard is aerial LiDAR flown from a helicopter or fixed wing aircraft. NHSRCL used it first on the 508 kilometre Mumbai-Ahmedabad High Speed Rail corridor, and reported completing the ground survey in about twelve weeks against the ten to twelve months a traditional survey would have taken. On the Delhi-Varanasi corridor the capture width was 300 metres, 150 metres either side of the proposed alignment, with 86 master control points and 350 secondary control points established against Survey of India benchmarks.
A quadcopter does not do that job. A 500 kilometre greenfield alignment through forest and hill terrain needs an aircraft-mounted laser scanner, and any provider quoting drone photogrammetry for it is quoting something that will not deliver.
What drone photogrammetry does well on railways is everything at project scale rather than network scale. Doubling and third line sections. Gauge conversion stretches. Station areas, yards and depots. Sidings into plants and ports. Railway land parcels and encroachment. Construction progress and earthwork on an awarded package. Structure documentation at bridges, ROBs and RUBs. These are the jobs where mobilising an aircraft makes no sense and a ground crew is slow, exposed to live traffic, or both.
Where a section is short but heavily wooded, drone-mounted LiDAR sits between the two. We will tell you which of the three your job needs before you engage us.
The survey stages, and what each one needs
Indian Railways runs project surveys through a defined sequence, set out in the Code for the Engineering Department and the Construction Manual. Knowing which stage you are at determines what the survey has to produce.
Reconnaissance survey. Broad corridor identification from maps and satellite data. Drone work rarely features here, and orthomosaic capture over specific problem areas, a river crossing, a built-up stretch, a forest edge, can settle a question that satellite imagery leaves open.
Preliminary engineering-cum-traffic survey. Topography, geology, hydrology, land use and traffic along the shortlisted corridors, feeding the techno-economic case. Corridor-scale terrain data matters here, and for shorter corridors drone photogrammetry produces it faster than a ground crew.
Final location survey. Precise final alignment and construction-level data. This is where accuracy requirements tighten and where the method question becomes serious. On open terrain, well-controlled drone photogrammetry produces the surface. Under canopy it does not, and the deliverable has to come from LiDAR or ground survey.
Execution and monitoring. Once the package is awarded, the survey question changes from what is the ground like to what has been built and how much material moved. This is where drone capture earns its keep month after month.
What we survey
Corridor sections. Doubling, third and fourth line, gauge conversion and short new line stretches. Existing formation levels, side drains, cross-drainage positions, adjoining land use and the boundary strip.
Stations, yards and depots. Track layout, platform geometry, circulating areas, parking, buildings, drainage and available space for redevelopment. Yard surveys are the case where a drone most obviously beats a ground crew, because the alternative is a surveyor walking between running lines.
Sidings and last-mile connectivity. Private sidings into cement plants, steel works, power stations, ports and logistics parks, where the alignment is short and the ground is congested.
Railway land and encroachment. Boundary strip mapping against the land plan, with structures, cultivation and occupation documented on a dated basis.
Structures. Bridges, road over bridges, road under bridges, retaining walls, tunnel portals and platform shelters, captured with oblique passes so vertical faces reconstruct properly.
Construction progress. Monthly capture on awarded packages, with earthwork quantities, structure progress and a dated visual record.
How we do it
Railway corridors combine the geometric problems of any linear survey with a set of operational constraints that no other sector has.
Step 1: Scoping and chainage
We work to your chainage from the start, because your alignment sheets, your land plan schedule and your bill of quantities are organised that way. A cross section at kilometre 118/4 needs to be findable at kilometre 118/4, and outputs referenced to arbitrary tile numbers create rework for your design team.
Corridor width is settled here. The boundary strip is one width, the earthwork corridor is another, and drainage catchment or borrow area mapping is wider still. On doubling work the width also has to cover the proposed second line, its formation and its drainage, which sits outside the existing land in places.
Coordinate system too. Most railway projects run on UTM WGS84, and a long east-west corridor can cross a zone boundary. Whether to hold one zone across the project or split at the boundary is decided before control goes in, not after.
Step 2: Two sets of permissions
Railway work needs DGCA clearance and railway administration clearance, and they are separate processes with separate lead times.
On the DGCA side we map the alignment against the Digital Sky airspace map and return the zone classification by chainage. Green zone sections can be flown to 120 metres above ground level without prior flight permission. Yellow zone sections need permission and carry a lower ceiling. Red zone sections cannot be flown. A long corridor routinely crosses all three, and it will also cross district and state boundaries that change who the coordinating authority is.
On the railway side, flying over operational railway land and installations needs permission from the railway administration, typically through the Divisional Railway Manager’s office and the relevant engineering department, arranged through your project authority. Some railway installations are treated as sensitive, and no amount of DGCA clearance substitutes for the railway’s own approval. We build this lead time into the programme rather than assuming it will arrive.
Our aircraft carry active Unique Identification Numbers and our pilots hold DGCA Remote Pilot Certificates. Long corridors are flown as segments with the crew relocating, because operation beyond visual line of sight needs specific DGCA permission.
Step 3: Control along the alignment
Corridors accumulate error along their length. A block survey is constrained on all four sides. A corridor is constrained only along its axis, which gives the bundle adjustment much less geometric strength, and the result drifts. The drift is smooth, which means it looks entirely plausible in the deliverable until somebody checks a coordinate at the far end.
We manage that with ground control spaced along the corridor at intervals set by the accuracy requirement, placed in staggered pairs on alternate sides of the alignment. Control in a single line down the centre leaves the block free to roll about that axis, and the resulting cross-slope error goes straight into your cross sections and your formation levels.
Where the project has existing railway benchmarks or master control points, we tie into them rather than establishing an independent frame, because a survey in its own coordinate system is of limited use to a design team working in the project’s.
Independent checkpoints go in along the length and are withheld from the adjustment. On long sections we report RMSE by segment as well as overall, since a single aggregate figure across 40 kilometres can conceal a bad stretch.
Step 4: Flight planning
Ground sample distance is pixel pitch multiplied by flying height divided by focal length. A 20 megapixel one-inch sensor with 5472 pixels across 13.2 millimetres has a pixel pitch of about 2.41 micrometres. With an 8.8 millimetre lens at the 120 metre ceiling that gives roughly 3.3 centimetres GSD, inside a 5 centimetre requirement with margin.
Corridor geometry sets the pattern. We plan at least three parallel strips across the corridor rather than a single line down the track, because a single strip carries no lateral overlap and reconstructs badly. Forward overlap runs at 80 percent, lateral at 70 percent or higher. Cross-strips go in perpendicular to the alignment at intervals, which is the standard control against longitudinal drift and the first thing a cheap quotation omits.
Sections with vertical variation, high embankments, deep cuttings, ghat sections and approach ramps to ROBs, are planned with terrain following so GSD holds across the full range.
Structures get separate treatment. Nadir imagery reconstructs the deck of a bridge and almost nothing of the piers. Bridges, ROBs, RUBs, retaining walls and portals are captured as small oblique blocks stitched into the corridor.
Step 5: Capture near live track
This is where railway work differs from every other corridor survey, and it is worth being specific about it.
Overhead equipment. Electrified sections carry 25 kV AC overhead equipment. Contact is catastrophic and proximity is dangerous without contact, because the conductor is thin, often invisible against a bright sky, and supported by portals and catenary wires that are harder to see than the contact wire itself. Every flight plan we produce marks OHE, portals and feeder crossings, and take-off, landing and transit paths are planned to keep clear of them at all times rather than relying on the pilot spotting them.
Magnetic interference. Steel rails, traction return current and OHE disturb a drone’s magnetometer. A compass calibrated on ballast next to a running line can be wrong. We calibrate away from track, monitor heading behaviour through the flight, and treat unexpected yaw drift as a reason to land rather than continue.
Live traffic. Nothing goes on or across the track. Take-off and landing points sit off railway land or in agreed locations clear of the running lines, and the crew works to whatever safety induction the railway administration requires. Where the work genuinely needs the drone over the running line, it is coordinated with the traffic block rather than squeezed between trains.
Moving trains in the imagery. A train passing through the block during capture produces ghosting in the mosaic and spurious points in the cloud. On busy sections we plan around the timetable where possible and filter what we cannot avoid.
Yard work. Station yards concentrate all of these problems in a small area, along with staff, passengers and road traffic in the circulating area. Yard surveys are planned with the station manager rather than turning up with a flight plan.
Step 6: Processing
Aerial triangulation with tie point extraction and bundle adjustment against the surveyed control, with camera self-calibration. On corridors we watch the distortion model closely, because uncorrected lens distortion on a long strip produces a systematic vertical bow along the alignment, and on a railway that reads as a false gradient in your formation levels.
Dense matching generates the point cloud. Classification then separates ground from vegetation, structures, plant and rolling stock. The hard cases on railway corridors are ballast shoulders, cess and side drains, all of which sit close to formation level, and the plantation strip along the boundary. These are checked by hand rather than accepted as the software produced them.
The DTM is gridded, imagery orthorectified and mosaicked with seamlines routed away from structures, and contours generated at your interval.
Step 7: Chainage-referenced outputs
Cross sections are cut perpendicular to the alignment at your specified interval and labelled by chainage. Longitudinal sections follow the alignment. Feature lines for track centre, cess, toe, crest, drain invert, boundary and structure edges come out as separate CAD layers named to your project convention.
Quantities are reported by chainage segment so they reconcile against your bill of quantities, rather than arriving as a single corridor total nobody can check.
Step 8: Quality control and delivery
Residuals at the withheld checkpoints, RMSE reported horizontally and vertically, by segment on long sections. The report states control layout, checkpoints held back, processing parameters and software. Deliverables are organised by package and chainage range, with projection metadata embedded in every spatial file.
Accuracy on a railway corridor
Ground sample distance and accuracy are different quantities. GSD is the ground area one pixel covers. Accuracy is how close a coordinate is to the truth.
For a well-controlled photogrammetric block, horizontal RMSE typically lands between one and two times the GSD and vertical between two and three times. At 5 cm GSD that suggests roughly 5 to 10 centimetres horizontal and 10 to 15 centimetres vertical. The figures in your report come from your checkpoints rather than from that rule.
Two things make railway accuracy requirements distinctive.
The first is that vertical tolerance on rail formation is tighter than on most road work, and the error that matters is relative rather than absolute. A uniform 10 centimetre offset across an entire section is a nuisance. A 10 centimetre variation over 200 metres is a false gradient, and on a railway a false gradient in the survey becomes a real problem in the design. This is why segment-level RMSE reporting matters more here than anywhere else.
The second is that photogrammetry measures the top of the ballast, not the rail. Where you need rail level, top of rail, cant and versine, that is track survey work with a track geometry trolley or a total station, and drone data does not substitute for it. We survey the formation, the earthwork, the drainage and the surroundings. The permanent way is somebody else’s instrument.
Canopy is the other limit. Boundary plantation, avenue trees and forest sections prevent the ground being seen, so the DTM there is interpolation. We mark those areas rather than presenting them as measured.
Deliverables and output file formats
| Deliverable | What it is | Format |
|---|---|---|
| Corridor orthomosaic | Georeferenced image of the section, tiled by chainage range | TIFF (GeoTIFF), JPEG |
| Digital terrain model | Bare earth surface for formation design and earthwork | TIFF |
| Digital surface model | Everything captured, including OHE structures, buildings and vegetation | TIFF |
| Contours | Generated from the DTM at your specified interval | DXF, SHP |
| Cross sections | Perpendicular to alignment at your interval, labelled by chainage | DXF, DWG, PDF, XLSX |
| Longitudinal section | Existing ground profile along the alignment | DXF, DWG, PDF |
| 3D point cloud | Classified, RGB attributed, for design software import | LAS, LAZ |
| Object marking and feature extraction | Track centre lines, cess, toe and crest, drains, culverts, OHE mast positions, signals, level crossings, boundary walls, buildings | DXF, TIFF |
| Station and yard layout plan | Track layout, platforms, buildings, circulating area, drainage | DWG, DXF, PDF |
| Topographic map | Layered drawing at your specified scale with levels and features | DWG, DXF, PDF |
| Railway land and encroachment layer | Boundary strip against the land plan, with structures and occupation attributed | SHP, DXF, PDF |
| Earthwork and quantity report | Cut and fill by chainage segment against design or previous epoch, base surface stated | PDF, XLSX |
| Change detection surface | Difference between two epochs, cut and fill mapped as raster | TIFF, PDF |
| Structure documentation | Oblique capture and mesh of bridges, ROBs, RUBs, walls and portals | OBJ, PDF, JPEG |
| Progress video and photo set | Dated corridor flythrough and georeferenced stills at work fronts | MP4, JPEG |
| Thermal orthomosaic | Where specified, for roof, structure or electrical survey | TIFF, JPEG |
| RMSE and quality report | Residuals overall and by segment, control layout, parameters | |
| Flight log and survey record | Flights, heights, times, drone UIN, pilot licence number |
Raw imagery is handed over with the deliverables.
Railway land, encroachment and redevelopment
Indian Railways holds one of the largest land portfolios in the country, much of it in long narrow strips through built-up areas, and much of it under pressure. Documenting what is physically standing on a boundary strip is slow and confrontational on foot. From the air it takes an afternoon.
A boundary strip survey produces the orthomosaic, the land parcel overlay against the land plan, and an attributed layer of structures, cultivation and occupation inside the boundary, with areas computed. Because it is dated and georeferenced, it also establishes a baseline: a second survey a year later shows what has changed.
The same data supports station redevelopment and land monetisation work, where the question is how much usable area exists, where the constraints sit and what would have to move.
The survey documents physical occupation. It does not determine title or the legal boundary, which come from the land records and the railway’s own land plan.
Construction progress and quantities
On an awarded package the commercial case for drone survey is the quantity measurement rather than the imagery.
Earthwork on a railway formation is measured, certified and paid. The traditional method is cross sections at intervals with interpolation between them, and the interpolation error is invisible because there is no data between sections to contradict it. A surveyed surface removes the interpolation, and quantities can be recomputed at any section spacing afterwards because the surface exists.
A pre-construction baseline flight is the single most valuable capture on the project and the one most often skipped. Original ground level along the full section, measured before the first machine moves, is what every subsequent quantity is calculated against. Without it, quantities get argued from design sections and borrow records.
Monthly capture then gives embankment volumes as layers are placed, borrow area depletion, structure progress at bridges and ROBs, and a dated visual record. Where a quantity is challenged later, there is a surface to point at.
Where we work
We are based in Ahmedabad and mobilise to project sections nationally.
Gujarat. The Western Dedicated Freight Corridor through Palanpur, Mehsana, Ahmedabad, Vadodara, Bharuch and Surat. Mumbai-Ahmedabad High Speed Rail sections through Sabarmati, Ahmedabad, Anand, Vadodara, Bharuch, Surat and Vapi. Ahmedabad Metro and the Sabarmati multimodal hub area. Doubling and gauge conversion work across Saurashtra and Kutch, and sidings into the plants and ports at Mundra, Kandla, Jamnagar and Hazira.
Maharashtra and the west. Corridor and yard work around Mumbai, Pune, Nashik, Nagpur, Solapur and Chhatrapati Sambhajinagar, and the Konkan alignment.
Northern India. Delhi NCR, Jaipur, Jodhpur, Ludhiana, Ambala, Chandigarh, Lucknow, Kanpur, Varanasi, Prayagraj and Patna, including Eastern DFC sections and metro packages.
Central India. Bhopal, Itarsi, Jabalpur, Indore, Gwalior, Raipur and Bilaspur, with heavy freight corridor and siding work through the coal belt.
Eastern India. Kolkata, Bhubaneswar, Rourkela, Jharsuguda, Ranchi, Dhanbad, Bokaro, Jamshedpur and Guwahati, including port and steel plant connectivity.
Southern India. Chennai, Bengaluru, Hyderabad, Vijayawada, Visakhapatnam, Coimbatore, Madurai, Kochi and Mangaluru, covering doubling, electrification support and metro packages.
What we do not do
Long greenfield alignment surveys at DPR scale need aircraft-mounted LiDAR. Drone photogrammetry does not substitute for it, and on a forested or hill corridor it will not produce a usable bare earth surface at all.
Track geometry is not our work. Rail level, gauge, cant, twist and versine need a track recording trolley or total station on the permanent way. We survey the formation and the surroundings.
Signalling and OHE design survey needs measurements our sensors do not take. We can map mast positions and portal locations from imagery, and clearance verification to design tolerance is a different exercise.
Underground and tunnel interiors are outside what aerial photogrammetry reaches. We survey portals and approaches.
Setting out and stake-out remain total station work. We survey what exists.
Sections in a red zone, and any railway installation the administration declines to permit, cannot be flown by anyone commercially.
Frequently asked questions
Can you fly over a live railway line?
Only with permission from the railway administration in addition to DGCA clearance, and with the flight planned around the overhead equipment and the traffic. Take-off and landing happen clear of the running lines, and where the survey genuinely needs the drone over the track it is coordinated with a traffic block rather than fitted between services.
Is drone photogrammetry acceptable for a final location survey?
It depends on the terrain and the specification. On open ground with proper control it produces the surface. Under canopy it does not, and the requirement has to be met with LiDAR or ground survey. Send us the survey specification and we will tell you plainly which applies.
Why did the bullet train corridors use LiDAR rather than drones?
Scale and canopy. A 500 kilometre alignment through forest, hills and dense settlement needs a sensor that penetrates vegetation and a platform that covers hundreds of kilometres a day. Aerial LiDAR was used first in India for a railway project on the Mumbai-Ahmedabad corridor, and NHSRCL reported completing the ground survey in around twelve weeks against ten to twelve months by traditional methods.
How accurate is a drone railway survey?
With correctly spaced staggered control, expect horizontal RMSE around one to two times the GSD and vertical around two to three times. At 5 cm GSD that is roughly 5 to 10 centimetres horizontal and 10 to 15 centimetres vertical. We report measured figures from withheld checkpoints, by segment as well as overall.
Will the survey give me rail levels?
No. Photogrammetry measures the top of ballast and surrounding ground. Rail level and track geometry need instrumentation on the permanent way.
How much corridor can you cover in a day?
It depends on corridor width, required GSD, airspace and access to take-off points along the alignment. We return a chainage-by-chainage plan rather than a single kilometres-per-day rate, because railway corridors vary too much along their length for an average to be useful.
Can you survey a station yard while it is operating?
Yes, with the railway administration’s permission and the station manager’s involvement in planning. Yards are the strongest case for drone survey precisely because the alternative puts a surveyor between running lines.
What file formats do you deliver?
Orthomosaic as GeoTIFF and JPEG. DTM and DSM as GeoTIFF. Contours as DXF and SHP. Cross sections, layout plans and topographic maps as DWG, DXF, PDF and XLSX. Point cloud as LAS and LAZ. Object marking as DXF and TIFF. Land and encroachment layers as SHP. Quantity reports as PDF and XLSX.
Can you document encroachment on railway land?
Yes. We produce a dated orthomosaic, the boundary strip against your land plan, and an attributed layer of structures and occupation with areas computed. The survey records what is physically there. It does not determine title or the legal boundary.
Do you handle the permissions?
We handle DGCA airspace permissions and return the zone classification by chainage. Railway administration permission has to come through your project authority, and we will tell you what to apply for and how long to allow.
Get a quote for your section
Send us the chainage range, the corridor width, the alignment as a KML or shapefile, the deliverables and whether the capture is one-off or recurring. Tell us whether the section is electrified and whether it is operational, because both change the plan. We will return the airspace classification by chainage, the control plan, the permissions you will need from the railway administration, a delivery schedule and a fixed price.