Drone City Survey and Urban Mapping in India

Geocartis carries out drone surveys of towns and cities for municipal corporations, urban development authorities and their consultants across India. We produce the georeferenced base map that everything else sits on: orthomosaic, building footprints, road network, drainage, open space and municipal assets, at 5 cm ground sample distance or better with DGPS control and an RMSE report. Outputs are delivered as GIS layers your team can load directly. 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 GIS software.

“City survey” means two different things, and the difference matters

Before anything else, a point of vocabulary that causes real confusion in Gujarat and Maharashtra.

In the revenue sense, city survey is a formal statutory process. A city survey office, under the state revenue department, surveys urban land, assigns city survey numbers and issues property cards recording rights. That is a legal determination of title and boundary, carried out by revenue officials under the land revenue code. No private agency does it, and no drone produces it.

In the general sense, city survey means measuring the physical city: what is built, where it is, how big it is, what condition it is in. That is a mapping exercise, and it is what we do.

The two are related in one direction only. Our survey shows what is physically standing on the ground, to a stated accuracy, on a stated date. It can be overlaid on the city survey records and it will often reveal that the two disagree. What it cannot do is settle which one is right. That is for the revenue authority, the city survey superintendent and, if it comes to it, a court.

We say this plainly on the first screen because urban clients occasionally arrive expecting a drone to resolve a title question, and it never will.

What urban local bodies actually use drone survey for

The GIS base map. The georeferenced foundation layer that every other municipal dataset registers to. Without one, the water network sits in one coordinate system, the property register in another and the road inventory in a spreadsheet.

Property tax survey. Building footprints with measured plan area, floor count from oblique imagery and shadow, usage indication, and linkage to the existing assessment register. This is the application that pays for everything else, and it gets its own section below.

Road and street inventory. Carriageway extent, surface type, footpath, kerb, median, street furniture, signage, poles, trees, and junction geometry, all measurable off the orthomosaic.

Storm water and drainage. Surface levels, natural flow direction, low points, catchment delineation and the existing drain network where it is visible. Terrain data at 5 cm GSD makes urban flooding analysis possible in a way that a contour map from the 1990s does not.

Municipal asset mapping. Parks, playgrounds, community halls, schools, health centres, water tanks, pumping stations, street lights, transformers, bins and public toilets, positioned and attributed.

Encroachment on municipal land. Plot boundaries against occupation, dated and georeferenced.

Slum and informal settlement mapping. Structure counts, footprint areas, access lane networks and open space, for redevelopment planning and beneficiary surveys.

Development plan and master plan support. Existing land use, built-up extent, plot coverage and the physical reality against which a proposed plan is drawn.

The scheme context, and what accuracy each needs

Two national programmes shape what urban clients ask for, and they sit at different resolutions.

AMRUT GIS base maps. MoHUA approved the sub-scheme on formulation of GIS-based master plans for 500 AMRUT cities in October 2015, as a fully centrally funded reform with a total cost of around ₹515 crore. It builds geo-referenced base maps and land use maps at 1:4000 scale, and the master plan is then formulated on that base under the state town and country planning act. AMRUT 2.0 extended the same approach to Class II towns with populations between 50,000 and 99,999, taking the total towards a thousand cities. MoHUA’s position is that states can source the base map either from National Remote Sensing Centre satellite imagery or from a drone survey.

That choice is the whole question. At 1:4000, satellite imagery is adequate for a master plan. It is not adequate for anything that touches an individual property.

SVAMITVA as the accuracy benchmark. The Ministry of Panchayati Raj’s SVAMITVA scheme maps inhabited rural abadi areas so that property cards can be issued. Survey of India carries out the drone-based large-scale mapping using survey-grade drones and the CORS network, and the stated target is positional accuracy of up to 5 centimetres for abadi properties, with mapping at scales as large as 1:500.

SVAMITVA is executed by Survey of India in rural areas, and it is not work a private agency bids for. What it establishes is the standard. When the government maps property parcels for the purpose of recording rights, it maps them at 5 centimetres, not at 1:4000. Any urban property-level exercise that aims lower than that is aiming at the wrong target.

Our default for urban work is 5 cm GSD or better, which supports mapping at 1:500 to 1:1000 and is where building-level measurement becomes defensible. Where a client only needs a 1:4000 master plan base, we say so and quote accordingly rather than selling resolution nobody will use.

Property tax survey, and the arithmetic behind it

For most municipal corporations the business case for a drone survey is property tax, and it is usually a strong one.

Assessment registers in Indian cities drift out of date in three ways at once. Properties are built and never assessed. Assessed properties add floors that are never recorded. And recorded plan areas were measured by somebody with a tape, decades ago, sometimes generously.

A drone survey at 5 cm GSD with oblique coverage addresses all three. Every structure in the survey area appears in the orthomosaic, so unassessed properties surface by comparing the footprint layer against the register. Floor count comes from oblique imagery, building height from the surface model, and the combination flags properties whose recorded floor count is short. And plan area is measured from the footprint rather than reported.

We are careful about what that measurement is. What we produce is roof footprint area, measured to a stated accuracy. Built-up area, carpet area and the assessable area under your specific municipal act are different quantities derived from it by rules that are yours, not ours. Roof overhangs, chajjas, balconies, parapets and covered parking all sit somewhere in the gap. We deliver the measured footprint and the height, we document the measurement basis, and your assessment team applies the rules. A provider who hands a municipal corporation a column labelled “taxable area” is inviting an appeal.

The survey also does not establish ownership, occupancy or usage. It shows that a structure exists, its size and its apparent character. Who owns it, who occupies it and what it is used for come from field verification. The value of the aerial survey is that it tells your field teams exactly which properties to visit, instead of sending them ward by ward to check everything.

How we do it

Urban survey is the most constrained work we do. The flying is technically straightforward and everything around it is difficult.

Step 1: Scoping by ward and zone

We scope urban work to your administrative structure rather than to convenient rectangles. Ward boundaries, zone boundaries and revenue village boundaries within the municipal limit all matter, because your data, your budgets and your field teams are organised that way. A deliverable that cuts across three wards is a deliverable somebody has to re-cut.

Coordinate system is settled at the start. Most municipal GIS runs on UTM WGS84, and some corporations hold a local grid inherited from an earlier survey. If your existing layers sit on that grid, we need the transformation parameters before control goes in, because a beautiful new base map that does not register to the existing water network is worse than no base map.

City limits change. Where an area has recently been merged into the corporation, the boundary you have and the boundary that is notified may differ, and that needs resolving before the flight plan is drawn.

Step 2: Airspace over a city

This is the hardest airspace problem in any sector we work in, and it is the reason urban projects need a longer lead time than clients expect.

Most Indian cities have an airport inside or adjacent to the built-up area, which puts a large share of the municipal limit inside the yellow zone extending 8 to 12 kilometres from the airport perimeter. In Ahmedabad, Sardar Vallabhbhai Patel International Airport sits within the city, so much of the northern and eastern city, including Hansol, Naroda, Kubernagar, Shahibaug, Airport Road and parts of Chandkheda, needs airspace permission and carries a ceiling lower than the 120 metres available in a green zone. The same pattern applies in Mumbai, Chennai, Kolkata, Hyderabad, Pune, Lucknow and Patna.

Beyond the airport, cities contain defence establishments, government complexes, secretariats, raj bhavans, high courts, central prisons and other installations around which flying is restricted or refused. A city-wide survey is therefore never a single permission. It is a mosaic, and part of it will be unflyable.

We map the entire survey area against the Digital Sky airspace map before quoting, and return the zone classification ward by ward with the restricted pockets marked. Where an area cannot be flown, you know at proposal stage and can decide whether to fill it from satellite imagery, a ground survey or not at all. Discovering it after mobilisation is expensive.

The Drone Rules also restrict operation over assemblies of people, which in an Indian city means the flight programme has to work around market days, festivals, processions and school hours.

Our aircraft carry active Unique Identification Numbers and our pilots hold DGCA Remote Pilot Certificates. Local police intimation is usually advisable even where it is not formally required, and we arrange it through the corporation.

Step 3: Privacy and data handling

An urban survey photographs private property. Rooftops, terraces, courtyards, private open space and, unavoidably, people. This deserves more than a line in a contract.

Rule 17 of the UAS Rules 2021 places the obligation on the operator not to share data gathered during operations with any third party without permission of the person the data pertains to. We treat that as the floor. Raw imagery and processed outputs go to the commissioning authority and nowhere else. We do not retain urban imagery for portfolio use, we do not publish sample outputs from municipal projects without written consent, and we do not use city survey data to train anything.

Nadir imagery at survey altitude does not identify individuals in any meaningful way. Oblique imagery, which we need for floor counts and facade information, is closer and can. Where a client has a concern, we can process oblique imagery so that only the derived measurements are retained and the raw obliques are returned and deleted from our systems on handover. It is worth agreeing that before the survey rather than after somebody asks a question in a standing committee meeting.

Step 4: Ground control in a built-up area

Cities are hostile to GNSS. Tall buildings block satellites and reflect their signals, so a receiver in a narrow lane can report a confident position that is a metre or more wrong through multipath. This is the single most common cause of a poor urban survey, and it is invisible unless you check for it.

We site control points in open locations with clear sky view: road junctions, open grounds, park corners, wide chowks, rooftops of low buildings where access is available. We occupy for longer than we would in open country, we check for multipath in the observation residuals, and we re-observe anything that looks unstable rather than accepting the first fix.

Marked crosses are useful on open ground and impractical on a busy road, so in dense areas we use existing hard features that are unambiguously identifiable in the imagery: painted road markings, manhole rim centres, distinct corners of paving. These get photographed at ground level and documented so a future survey can reoccupy the same physical point.

Independent checkpoints go in across the area and are withheld from the bundle adjustment. On a survey covering a whole city we report RMSE by zone as well as overall, because the dense old core and the open peripheral wards behave very differently and a single city-wide figure hides that.

Step 5: 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. Inside a yellow zone with a lower ceiling the GSD improves but each photograph covers less ground, so image count and flight time rise sharply. Over a large municipal area that difference is measured in weeks.

Urban work needs oblique coverage as standard, not as an extra. A nadir-only block gives you roofs and nothing else. It cannot count floors, it cannot show a facade, it reconstructs building walls as vertical smears, and it systematically misplaces the base of tall buildings. Every urban survey we fly combines a nadir block with oblique passes at multiple headings.

Tall buildings create their own problems. A 60 metre tower flown at 120 metres leaves 60 metres of clearance, the GSD on its roof is half the GSD at street level, and the building occludes the ground around it in every image taken from one side. Areas with high-rise development get additional flight lines and tighter spacing so that every patch of ground is seen from several directions.

Overlap runs at 80 percent forward and 70 percent lateral as a minimum, and higher in dense areas, because urban scenes with repetitive facades and strong shadows are hard for tie point matching.

Step 6: Capture

Middle of the day for sun angle, because in a dense city long shadows fill narrow lanes completely and the ground there simply is not visible in the imagery. This is the reason winter surveys in northern cities need careful scheduling: low sun angles produce shadows that never clear from north-facing lanes.

Take-off and landing sites in a city are a real constraint. They need to be clear of people, clear of overhead cables, accessible, and secure enough that a crowd does not form. On a municipal project the corporation’s ward office usually helps, and we plan sites rather than improvising.

Overhead cables are the operational hazard. Indian city streets carry electricity distribution, street lighting, cable television and internet lines at low height, often uncharted and frequently invisible against a bright sky. Transit and landing paths are planned to avoid them, and the crew treats an unfamiliar street as cabled until it can see otherwise.

Crowds gather. A drone in an Indian residential lane attracts an audience within a minute, and an audience is both a safety issue and a delay. We work with visible identification, with the corporation’s letter, and where the area warrants it with a ward-level intimation beforehand.

Step 7: Processing and feature extraction

Aerial triangulation with bundle adjustment against the control and camera self-calibration. Dense matching to generate the point cloud. Classification to separate ground, buildings and vegetation, which produces the terrain model, the surface model and the building layer.

Then the part that takes the time. An urban base map is not an image, it is a set of attributed vector layers, and producing those is interpretation work.

Building footprints are digitised from the orthomosaic and the point cloud, with height from the surface model and floor count from the obliques. Roads are digitised as centreline and edge. Drainage, footpath, kerb, median, open space, water bodies and vegetation each become layers. Point features, poles, trees, manholes, street lights, bins, transformers, are marked and attributed.

Automated building extraction gets you perhaps most of the way in a planned area and much less in a dense old core, where adjacent buildings share walls, roof levels step irregularly and the aerial view gives no clue where one property ends and the next begins. Those areas need manual digitisation checked against whatever cadastral or property register information exists, and sometimes against a field visit. We are explicit in the deliverable about which areas were extracted with confidence and which are interpretation.

Layer naming, attribute schema and coding follow your existing convention, or the AMRUT spatial data model where the project sits under that scheme. Delivering a technically excellent dataset in our own schema creates work for your GIS team rather than saving it.

Step 8: Quality control and delivery

Residuals at withheld checkpoints, RMSE reported horizontally and vertically, by zone as well as overall. Feature layers checked for topology: closed polygons, no gaps or overlaps in the footprint layer, connected road network. Attribute completeness checked against the agreed schema.

Delivery is organised by ward and zone, with projection metadata embedded in every spatial file, and in the format your GIS actually ingests.

Accuracy in an urban environment

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. Your report carries the measured figures from your checkpoints.

Three urban conditions push against that.

Multipath in the control observations, discussed above, is the big one, and it is why control siting matters more in a city than anywhere else.

Occlusion is the second. Ground in a narrow lane between four-storey buildings may be visible in very few images, or none. Where the ground is not seen, the surface there is interpolation, and we mark it rather than presenting it as measured.

Building footprint accuracy is a separate question from point accuracy, and clients often conflate them. A footprint digitised from a nadir orthomosaic is the roof outline, not the wall line at ground level. On a building with a significant overhang those differ by half a metre or more. Where wall-line accuracy matters, oblique imagery and the point cloud are used to derive it, and the deliverable states which line was digitised. For property tax work this distinction is not academic, because it is exactly the sort of thing an assessee’s advocate will raise.

Deliverables and output file formats

Raw imagery is handed over on completion, subject to the data handling terms agreed at the outset.

The recurring model for a municipal corporation

A one-off city survey is a snapshot that starts going out of date the week it is delivered. Indian cities change fast enough that a five-year-old base map is materially wrong.

The pattern that works is a baseline survey followed by periodic re-capture, and it produces something a single survey cannot: change.

Re-flying an area annually or every two years, registered to the same permanent control, shows new construction since the last epoch, additional floors added to existing buildings, new encroachment on municipal land and open space, and changes to the road network. For a property tax department, the change layer is more immediately useful than the base map, because it is a directed list of properties to inspect rather than a whole-city dataset to work through.

It also means the base map stays current for every other department that depends on it, rather than degrading until somebody funds a fresh survey from scratch.

For budgeting, this converts an occasional large capital purchase into a defined annual programme, which is usually easier to sustain through a municipal budget cycle than a one-time item that competes with road works.

Other urban applications

Development plan and town planning schemes. Existing ground reality against the proposed plan, plot coverage, road width verification and the physical survey supporting a TP scheme, which in Gujarat is a substantial and continuous workload.

Storm water and flood management. Terrain modelling, catchment delineation, low-point identification and flood inundation visualisation built on the DTM. Cities with recurring monsoon flooding get more value from this than from almost anything else in the list.

Water and sewerage network planning. Surface levels along proposed alignments, existing chamber and manhole positions, and terrain for gravity main design.

Solid waste. Bin locations, collection route planning, transfer station layouts, and dumpsite volumes where the corporation has a legacy waste site.

Heritage and conservation. Documentation of heritage precincts and individual structures, which in Ahmedabad has particular relevance given the walled city’s World Heritage inscription.

Disaster response. Rapid post-event capture after flooding, fire or building collapse, giving an incident commander a current picture within hours.

Election and civic administration. Ward delimitation support, polling station catchment mapping and route planning.

What we do not do

We do not determine title, ownership or legal boundaries. City survey records, property cards and cadastral determinations come from the revenue department and the city survey office.

We do not produce assessable or taxable area. We measure footprint and height and document the basis. Applying your municipal act’s rules to reach an assessable figure is your assessment team’s function.

We do not carry out SVAMITVA mapping. That is executed by Survey of India in rural abadi areas.

We do not survey building interiors, and we do not measure carpet area.

We do not fly in red zones or over restricted installations, and we do not fly over assemblies of people.

We do not carry out the field verification that turns a footprint layer into an assessment. We can plan and support it, and somebody has to knock on the door.

Where we work

We are based in Ahmedabad and work with corporations, municipalities and development authorities nationally.

Gujarat. Ahmedabad, Gandhinagar, Surat, Vadodara, Rajkot, Bhavnagar, Jamnagar, Junagadh, Anand, Nadiad, Mehsana, Morbi, Bharuch, Navsari, Valsad, Bhuj and Gandhidham, plus the urban development authorities and the many Class II towns brought into AMRUT 2.0.

Maharashtra. Mumbai, Navi Mumbai, Thane, Pune, Pimpri-Chinchwad, Nashik, Nagpur, Chhatrapati Sambhajinagar, Solapur and Kolhapur.

Rajasthan and the north. Jaipur, Jodhpur, Udaipur, Kota, Ajmer, Bikaner, plus Delhi NCR including Gurugram, Noida, Ghaziabad and Faridabad, and Chandigarh, Ludhiana, Amritsar, Dehradun, Lucknow, Kanpur, Agra, Varanasi, Prayagraj and Patna.

Central India. Indore, Bhopal, Jabalpur, Gwalior, Ujjain, Raipur and Bilaspur.

Eastern India. Kolkata, Howrah, Bhubaneswar, Cuttack, Rourkela, Ranchi, Jamshedpur, Dhanbad and Guwahati.

Southern India. Hyderabad, Bengaluru, Chennai, Coimbatore, Madurai, Tiruchirappalli, Vijayawada, Visakhapatnam, Kochi, Thiruvananthapuram, Mysuru and Mangaluru.

Frequently asked questions

Can a drone survey replace our city survey records? No. City survey records are a legal determination of rights carried out by the revenue department. Our survey records what is physically on the ground and can be overlaid on those records, which frequently reveals discrepancies. Resolving a discrepancy is a revenue process.

Can you find unassessed properties for property tax? Yes, by comparing the surveyed building footprint layer against your assessment register. The survey identifies structures that do not appear in the register and structures whose recorded floor count or plan area looks short. Field verification then confirms each case.

Will you give us the taxable area of each property? We give measured roof footprint area and building height with the measurement basis documented. Converting that to assessable area under your municipal act involves rules about overhangs, balconies, parapets and usage that are yours to apply. We deliver the measurement, not the assessment.

What resolution do we need? For an AMRUT master plan base map at 1:4000, satellite imagery may be sufficient and MoHUA permits it. For anything at property level, 5 cm GSD or better, which supports 1:500 to 1:1000. As a reference point, Survey of India’s target under SVAMITVA for mapping abadi property parcels is positional accuracy of up to 5 centimetres.

Can you survey the whole municipal area? Usually most of it. Airspace restrictions around airports, defence establishments and government complexes will leave pockets that cannot be flown. We return the zone classification ward by ward at proposal stage so you know exactly which areas are affected before committing.

How do you handle privacy? Rule 17 of the UAS Rules 2021 prohibits sharing survey data with third parties without permission. Imagery goes to the commissioning authority only. We do not publish samples from municipal projects without written consent, and where oblique imagery is a concern we can return and delete the raw obliques on handover, retaining only the derived measurements.

How long does a city survey take? It depends on area, density, airspace and how much of the work is feature extraction rather than flying. Flying a mid-sized municipal area is a matter of weeks. Digitising and attributing the layers is usually the longer half of the programme, and we quote the two separately so you can see where the time goes.

Can you match our existing GIS? Yes. Tell us the coordinate system, the layer naming and the attribute schema you use, and we deliver into it. Where the project sits under AMRUT, we work to the scheme’s spatial data model.

What file formats do you deliver? Orthomosaic as GeoTIFF and JPEG. DTM and DSM as GeoTIFF. Vector layers as SHP, GeoJSON and DXF. Point cloud as LAS and LAZ. Base map sheets as PDF and DWG. 3D city model as OBJ or SHP with height attributes. Reports as PDF.

Can we compare against a survey done a few years ago? Depends on whether the earlier survey’s control was properly recorded. Where it was, we can register to the same frame and produce a genuine change layer. Where it was not, the two datasets can be overlaid but apparent differences may be survey error rather than real change, and we will say so rather than presenting a change layer that is partly artefact.

Get a quote for your city

Send us the survey area as a KML or shapefile with the ward or zone boundaries, the population or approximate built-up area, the deliverables and the coordinate system your GIS uses. Tell us whether the work sits under AMRUT, a property tax exercise, a development plan or something else, and whether you need oblique imagery.

We will return the airspace classification by ward with the restricted pockets marked, the control plan, a split of flying time against feature extraction time, a delivery schedule and a fixed price.