Drone Canal and Irrigation Survey Services in India

Geocartis surveys canal systems across India for irrigation departments, water resources departments, project authorities and their consultants. We cover alignment and topographic survey for new canals, bank and berm profiles, cross sections at chainage, silt and lining condition assessment, encroachment on canal land, and command area mapping. Capture 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 work nationally.

We are a survey and geospatial services company. We do not build drones and we do not sell software.

Canals are a different survey problem, in three specific ways

Canal work borrows from corridor survey and then adds constraints that no road or railway has.

They are extremely long relative to their width. A minor canal is perhaps six metres across and forty kilometres long. That is a survey with almost no lateral extent, which is the worst possible geometry for a photogrammetric block and the reason canal surveys drift more than any other corridor work if the control is not designed properly.

They contain water, and photogrammetry cannot see through it. This is the constraint that determines the whole programme, and it gets its own section below.

They are seasonal. A canal that is running cannot be surveyed for bed level. A canal in closure can. The window is fixed by the irrigation calendar, not by your project schedule, and if you miss it the next one is a year away.

Anybody quoting you a canal survey without asking when the canal is closed has not thought about the job.

The water problem and the closure window

A camera records reflected light from a surface. On water it records the surface, or a confusion of sun glint and sky reflection, and nothing below. There is no processing step that recovers bed level from an aerial photograph of a running canal.

This has hard consequences.

Silt measurement requires a dry canal. If you want to know how much silt has accumulated in a reach, you need bed level, which means surveying during the closure period. A survey of a running canal tells you the water surface elevation and the top of the bank, which are useful for other purposes and useless for desilting quantities.

Lining condition assessment requires a dry canal. Cracks, spalling, displaced panels, joint failures and undermining below the water line are visible only when the water is out.

Bed profile and L-section require a dry canal. The longitudinal bed profile that any rehabilitation design depends on cannot be flown while the canal is in operation.

Everything else can be done wet. Bank and berm geometry, service road condition, encroachment, land use, canal-top structures, cross drainage works above water level, command area mapping and vegetation on the banks are all measurable with water in the channel.

So the first question we ask on any canal enquiry is what you actually need, and the second is when the reach is closed. On a large system the closure is usually a defined annual maintenance period, often a few weeks, and it is when every desilting and lining contractor in the state also wants surveys done. Booking that window early matters more than negotiating the last five percent on price.

Where bed level is needed and the canal cannot be emptied, the answer is bathymetry with an echo sounder from a boat or a remote vessel, merged with the aerial survey of everything above water. That is a different scope with a different provider, and we will say so rather than delivering an interpolated bed and letting somebody design against it.

Where drone survey fits across a canal project

Alignment and feasibility. Terrain along candidate alignments, existing land use, settlements, crossings and structures. Canals are gravity systems, so alignment is driven by contour more tightly than any road, and a good terrain model at the option stage saves a lot of argument later.

Detailed project report. Topographic survey along the alignment, L-section, cross sections at chainage, cross drainage locations and the land to be acquired. Where the corridor is heavily wooded, the honest position is the same one as on our road and railway work: photogrammetry does not see through canopy, and either LiDAR or a ground crew has to fill those reaches.

Land acquisition. Corridor mapping against revenue boundaries, with structures, trees and crops documented on a dated basis inside the proposed acquisition strip.

Pre-construction baseline. Original ground level along the full alignment before excavation begins. Every earthwork quantity on the project is measured against it, and skipping it is how quantity disputes start.

Construction monitoring. Monthly or stage capture with excavation and embankment volumes by chainage segment, lining progress, and structure progress at regulators, escapes, aqueducts and siphons.

As-built. Final geometry against design, with the bed profile and section survey done before water is let in, because after that the opportunity is gone for a year.

Operation and maintenance. Silt accumulation, lining condition, seepage, encroachment, bank erosion, service road condition and vegetation. This is the recurring work, and on an established system it is most of the market.

How we do it

Step 1: Scoping, chainage and the closure calendar

We work to your chainage. Canal systems are referenced by chainage from the head regulator, and every deliverable we produce carries that reference so it matches your L-section, your bill of quantities and your inspection records.

Scoping settles the corridor width, which on canal work is more variable than people expect. A bed and bank survey needs perhaps thirty metres. An encroachment survey needs the full canal land boundary, which may be sixty or a hundred metres. A command area survey is measured in square kilometres. Quoting these as one job produces a number nobody can use.

Then the calendar. We need to know the closure period for the reach, whether it is a full closure or a rotational one, and how firm the dates are. On systems where closure is decided close to the date, we plan for a mobilisation window rather than a fixed day.

Coordinate system is settled at the outset. Most irrigation departments work in UTM WGS84, some in a project datum inherited from the original survey of the system, which on older canals may date back decades. If your L-section is in a local datum, we need the transformation before control goes in.

Step 2: Control along a very narrow corridor

Canals have the weakest geometry of any corridor we survey. A road has carriageway, shoulders, drains and a right of way that together give a usable block width. A minor canal with a service road on one side is a strip a few tens of metres across running for kilometres.

Three things follow.

Control spacing along the alignment is tighter than on wider corridors, because there is less lateral extent to constrain the adjustment.

Control must be staggered on alternate sides of the canal, never in a single line along one bank. A single line leaves the block free to roll about that axis, and a roll error on a canal survey is a cross-slope error in your section, which on a gravity system is the one error that matters most.

Cross-strips are mandatory rather than optional. On a narrow corridor they are the only thing preventing longitudinal drift, and drift on a canal L-section produces a false gradient. A canal designed against a false gradient does not carry the design discharge.

Where the system has existing benchmarks, we tie into them. An irrigation department’s L-section is a document with decades of history, and a new survey in an independent datum that disagrees with it creates a problem rather than solving one.

Independent checkpoints go in along the length and are withheld from the adjustment. We report RMSE by reach as well as overall, because a forty kilometre survey with a single aggregate figure can hide a bad section entirely.

Step 3: Airspace along the alignment

A canal crosses whatever is in its path. Over forty kilometres that can mean a green zone at the head, a yellow zone where the alignment passes within twelve kilometres of an airfield, district and sometimes state boundaries, and occasionally a restricted installation.

We map the whole alignment against the Digital Sky airspace map before quoting and return the zone classification by chainage. Where a reach cannot be flown you know at proposal stage.

Canals also run alongside and under other infrastructure. The Gujarat Chief Minister’s assembly reply in March 2025 on pending Narmada works noted that remaining minor canal construction depends on land acquisition and permissions from various departments because the alignments pass under utilities such as oil pipelines and power lines. Those same crossings are flight hazards, and every corridor plan we produce marks transmission line crossings as no-fly bands at operating altitude.

Long alignments are flown as segments with the crew relocating, because operation beyond visual line of sight needs specific DGCA permission. Our aircraft carry active Unique Identification Numbers and our pilots hold DGCA Remote Pilot Certificates.

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 green zone ceiling that gives roughly 3.3 centimetres GSD.

For canal work we routinely fly lower than the ceiling. A narrow channel needs enough pixels across the section to resolve bed and bank detail, and on a small minor a 3 cm GSD leaves the whole channel only a couple of hundred pixels wide. Where lining condition is the deliverable, we fly lower still, because a hairline crack in concrete lining needs sub-centimetre resolution to be identifiable at all.

Flight pattern is a minimum of three parallel strips across the corridor with cross-strips at intervals. Overlap at 80 percent forward and 70 percent lateral or better.

Lined canals present a specific matching problem. A long run of clean concrete lining is a low-texture surface with a strongly repetitive pattern, which is close to the worst case for tie point extraction. Panels look identical to each other, and the matching algorithm can associate the wrong ones. We manage that with higher overlap through lined reaches, and by placing additional temporary targets along the lining where a reach is particularly featureless.

Structures get oblique passes. Regulators, head works, escapes, aqueducts, siphons, falls and cross drainage works are vertical structures that nadir imagery reconstructs badly, and they are usually the parts of the system somebody actually wants to look at.

Step 5: Capture

Sun angle matters more on a canal than on flat ground, because a canal is a trench. A deep section in low sun is a trench full of shadow, and the shadowed bank face produces a hole in the point cloud exactly where the section needs data. Middle of the day, and where the canal runs east to west the timing window is narrower still.

Water, where present, is masked. We map the water surface extent as a polygon and exclude it from the terrain model rather than letting the matching algorithm generate noise across it. Wet patches, puddles in a closed canal and standing water in the bed all get the same treatment, and on a closure survey they are worth pumping out beforehand if the bed level there matters.

Access along a canal is usually good, because most systems have an inspection road on at least one bank. That is one of the few things that makes canal work easier than other corridor survey, and it means take-off points can be spaced along the alignment without difficulty.

The crew keeps a flight log covering flights, heights, times, control occupation and any anomaly, and that log forms part of the handover.

Step 6: Processing

Aerial triangulation with bundle adjustment against the surveyed control and camera self-calibration. On narrow corridors the distortion model needs particular attention, because uncorrected lens distortion on a long thin strip produces a systematic vertical bow, and a vertical bow along a canal reads as a gradient that is not there.

Dense matching, then classification. On canal corridors the hard cases are bank vegetation, which on older systems can be substantial, and the berm and toe lines where the classifier has to decide what is engineered surface and what is grown-over spoil. Silt in the bed of a closed canal is ground for the purposes of a bed level survey and is the thing being measured for a desilting quantity, so the deliverable has to be clear about which surface is which.

Step 7: Chainage-referenced outputs

Cross sections at your specified interval, cut perpendicular to the alignment and labelled by chainage. Longitudinal section along the bed, or along the design bed line where the bed is silted, with both shown where a desilting quantity is required.

Feature lines for bed edge, toe, berm, top of bank, lining edge, service road edge and canal land boundary, extracted as separate CAD layers named to your project convention.

Quantities reported by chainage segment so they reconcile against your bill of quantities rather than arriving as one figure for the whole reach.

Step 8: Quality control and delivery

Residuals at withheld checkpoints, RMSE reported horizontally and vertically, by reach as well as overall. Deliverables organised by reach and chainage range, with projection metadata embedded in every spatial file.

Accuracy on a canal

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 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. Your figures come from your checkpoints.

Canal work has a particular sensitivity that is worth stating plainly. Irrigation canals run at very flat gradients, often in the range of one in five thousand to one in ten thousand. At one in ten thousand, the design fall over a kilometre is ten centimetres, which is the same order as the vertical RMSE of a good photogrammetric survey.

That does not make drone survey useless for canals. It does mean two things. First, relative accuracy over short distances matters more than absolute accuracy, and it is generally much better than the absolute figure. Second, where a design depends on the fall over a reach, photogrammetric levels are not a substitute for precise levelling along the bed. We survey the geometry, the section, the volumes and the condition. Setting the design gradient on a flat-gradient canal is levelling work.

We say this on the page rather than in a footnote after the survey, because it is the one place where an over-sold drone survey can do real damage to an irrigation project.

Two site conditions degrade results. Bank vegetation prevents the ground being seen, so the surface there is interpolation and we mark it. Water, as discussed, is excluded rather than guessed.

Deliverables and output file formats

Raw imagery is handed over with the deliverables.

Operation and maintenance work

On an established system the recurring survey work is worth more than the construction work, and it clusters around five questions.

How much silt is in it. Bed level against design bed level, by reach, giving a desilting quantity that can be tendered and paid against a measurement rather than an estimate. Requires closure.

Is the lining failing. Low-altitude capture along a dry reach, with defects located by chainage and side and classified by type. This produces a prioritised rehabilitation list instead of a walking inspection that covers whichever reaches somebody had time for.

Where is it leaking. Thermal imagery can indicate seepage where water emerging on a bank or berm produces a temperature contrast against the surrounding surface. It works best in the right conditions, which usually means early morning or after dark when the thermal contrast is strongest, and it indicates rather than proves. A thermal anomaly tells your engineer where to investigate. It does not measure a seepage rate.

Who has built on the canal land. Canal land is a long narrow strip through agricultural and increasingly peri-urban land, and it attracts encroachment for the same reasons railway land does. A dated aerial survey against the canal land boundary produces the record, and a repeat survey produces the change.

Is the service road usable. Inspection roads along canal banks are how the system is maintained, and their condition is measurable off the same dataset.

Command area mapping

Beyond the channel itself, the command is where the project’s purpose is measured.

We map the culturable command area boundary, the extent actually under irrigation, cropping pattern indication from imagery, field channel networks where visible, and the drainage that carries water away. Repeat capture across a season or across years shows whether the irrigated extent is expanding as the distribution network is completed.

This is the dataset that answers the question every irrigation project eventually faces, which is whether water reaching the head of the system is reaching the fields. Canal capacity is a design figure. Irrigated extent is an observation.

Where we work

We are based in Ahmedabad, which puts us alongside the largest canal network in the country.

Gujarat and the Narmada system. The Sardar Sarovar Project’s main canal runs 458 km within Gujarat and continues into Rajasthan, with a design capacity of 40,000 cusec at the head, making it the largest irrigation canal in the world by carrying capacity. Below it sits a branch, distributary, minor and sub-minor network planned at over 70,000 km. As of the state government’s position in early 2025, around 5,900 km of that network remained to be constructed, including roughly 46 km of branch canals, 160 km of distributaries, 1,052 km of minors and 4,663 km of sub-minors.

That means two distinct markets in one state. New construction on thousands of kilometres of minor and sub-minor canal, and operation and maintenance survey on tens of thousands of kilometres already running. We work across the system from Kevadia and Bharuch through Vadodara, Anand, Kheda, Ahmedabad, Mehsana, Patan, Banaskantha, Surendranagar, Rajkot, Morbi and into Kutch, alongside the older Mahi, Ukai-Kakrapar and Shetrunji commands.

Rajasthan. The Indira Gandhi Canal through Sri Ganganagar, Hanumangarh, Bikaner and Jaisalmer, the Narmada canal command in Barmer and Jalore, and the Chambal and Mahi Bajaj Sagar commands.

Punjab, Haryana and the north. Bhakra and Sirhind systems, the Western Yamuna Canal, and the Upper and Lower Ganga Canal and Sharda Sahayak systems through Uttar Pradesh, with access via Chandigarh, Ludhiana, Meerut, Aligarh, Kanpur and Lucknow.

Central India. Chambal, Bargi, Indira Sagar and Omkareshwar commands, with access via Bhopal, Indore, Jabalpur and Gwalior.

Maharashtra. Jayakwadi, Ujani, Gosikhurd and the Krishna and Godavari valley systems, via Pune, Nashik, Chhatrapati Sambhajinagar, Solapur and Nagpur.

Southern India. Nagarjuna Sagar and Kaleshwaram in Telangana, the Godavari and Krishna delta systems and Polavaram command in Andhra Pradesh, Tungabhadra and Upper Krishna in Karnataka, and the Cauvery delta in Tamil Nadu, with access via Hyderabad, Vijayawada, Visakhapatnam, Bengaluru, Kalaburagi, Chennai and Tiruchirappalli.

Eastern India. Hirakud command in Odisha, Sone and Gandak systems in Bihar, and the Damodar Valley canals in West Bengal, via Bhubaneswar, Sambalpur, Patna and Kolkata.

What we do not do

We do not measure below water. A running canal cannot be surveyed for bed level by any aerial camera. Where bed level is needed and closure is not possible, bathymetry with an echo sounder is the method, and it is a separate scope.

We do not set design gradients on flat-gradient canals. Precise levelling along the bed is the correct instrument where the design fall over a kilometre is comparable to the survey’s vertical accuracy.

We do not see through canopy. Reaches under substantial tree cover need LiDAR or a ground crew, and we will say which before you engage us.

We do not measure discharge, seepage rate or water quality. Thermal imagery indicates where to investigate seepage. It does not quantify it.

We do not determine the legal canal land boundary. We map what is physically there against the boundary your department provides. The boundary itself comes from the acquisition record and the revenue department.

We do not fly in red zones.

Frequently asked questions

Can you survey a canal that has water in it? Yes, for everything above the water line: banks, berms, service road, structures, encroachment, land use and command area. No, for bed level, silt quantity or lining condition below the water line. Those need the canal to be in closure.

When should we schedule the survey? For anything requiring bed level, during the annual closure or maintenance period for that reach. Book it early, because the closure window is short and every desilting and rehabilitation survey in the state competes for it. For work above water level, any time outside the monsoon.

Can you give us a desilting quantity? Yes, from a closure survey. We measure the current bed surface against the design bed level and report the volume by chainage segment with the base surface stated. Silt density and the conversion to tonnage, where a tender needs it, come from sampling rather than assumption.

How accurate is a drone canal survey? With properly spaced and 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, measured from withheld checkpoints and reported by reach.

Is that good enough for canal design? For section geometry, quantities, condition assessment and command mapping, yes. For setting the design gradient on a canal running at one in ten thousand, no. At that gradient the fall over a kilometre is ten centimetres, which is comparable to the survey’s vertical accuracy, and precise levelling is the right instrument. We are explicit about this at proposal stage.

Can you detect seepage? Thermal imagery can indicate seepage where emerging water produces a temperature contrast against the bank. It works best in high-contrast conditions and it locates rather than quantifies. Treat it as a directed inspection list for your engineers.

Can you map encroachment on canal land? Yes, provided you supply the canal land boundary. We produce a dated orthomosaic, the boundary overlay and an attributed layer of structures and occupation with areas computed. The survey records physical occupation. It does not determine title.

How much canal can you cover in a day? It depends on corridor width, required GSD and airspace. Bank-to-bank survey of a minor at standard resolution covers a good number of kilometres a day. Low-altitude lining condition survey covers far less, because the flying height is much lower. We quote the two separately.

What file formats do you deliver? Orthomosaic as GeoTIFF and JPEG. DTM and DSM as GeoTIFF. Contours as DXF and SHP. Cross sections and L-sections as DXF, DWG, PDF and XLSX. Point cloud as LAS and LAZ. Object marking as DXF and TIFF. Encroachment and command layers as SHP. Quantity reports as PDF and XLSX.

Can you work to our department’s existing L-section datum? Yes, given the transformation parameters. Many irrigation systems carry a datum from the original survey, and a new survey that disagrees with the department’s historical L-section causes more trouble than it solves. We tie in rather than starting fresh.

Get a quote for your reach

Send us the alignment as a KML or shapefile with the chainage range, the corridor width you need, whether the canal is lined, and the deliverables. Tell us whether the reach will be dry or running at the time of survey, and if dry, the closure dates.

We will return the airspace classification by chainage, the control plan, a delivery schedule and a fixed price. If bed level is part of the scope and the closure window is close, tell us now rather than in the closure month.