Drone Land Survey Services in India

Geocartis is a drone survey and geospatial services company based in Ahmedabad that carries out land surveys across India. We fly your site, process the imagery and hand over georeferenced data: orthomosaic, digital terrain model, contours, 3D point cloud and volume calculations. Standard capture resolution is 5 cm ground sample distance or better, ground control is established with DGPS, and an RMSE report goes out with every dataset.

We are a services company. We do not manufacture or sell drones, and we do not licence software. What we sell is the survey and the data that comes out of it, delivered in the formats your engineers, your CAD team or your regulator actually need.

What a drone land survey gives you that a ground survey does not

Every point a total station records is a point somebody walked to. On a quarry bench, a live construction deck or a 200 hectare industrial plot, that limits you to the points the surveyor judged worth shooting. The gaps between them get interpolated.

Aerial photogrammetry inverts that. At 5 cm ground sample distance, one hectare of ground is recorded as roughly four million pixels. The dense point cloud built from that imagery carries tens of thousands to a few hundred thousand points per hectare depending on the matching density we set. You are no longer choosing which points matter before you have seen the site. You capture the surface, then decide.

Three practical consequences follow.

The first is time on site. A block of 100 hectares flown at 5 cm GSD needs somewhere around 330 to 350 photographs at standard overlaps, which is one to two flight sessions depending on battery cycles and wind. A ground crew covering the same area to a comparable level of detail is a multi-day exercise.

The second is safety. Nobody walks the highwall, the ash pond bund, the tailings dam face, the roof of a transmission tower or the live carriageway of a national highway. The sensor goes where the surveyor should not.

The third is comparability. Two surveys of the same site flown to the same specification, over the same permanent ground control, can be differenced directly. That is what makes annual compliance surveys and monthly construction progress tracking work at all. Change detection between two point clouds is arithmetic. Change detection between two total station traverses is guesswork about whether the second crew shot the same features as the first.

Drone land survey in Ahmedabad

Ahmedabad is where we are based and where most of our short-notice work happens. Sites in and around the city fall into a few recognisable groups: industrial plots in Sanand, Changodar, Bavla and Kheda; warehousing and logistics parks along the Ahmedabad-Rajkot and Ahmedabad-Vadodara corridors; residential and commercial development across Shela, Bopal, Shantigram, Chandkheda and the SG Highway belt; and infrastructure work tied to the Ahmedabad-Dholera Expressway, the Ahmedabad-Mumbai high speed rail corridor and the Sabarmati riverfront extensions.

One local detail matters more than most clients expect. Sardar Vallabhbhai Patel International Airport sits inside the built-up area of the city, which means a large part of northern and eastern Ahmedabad falls inside the yellow zone on the DGCA airspace map. Sites in Hansol, Naroda, Kubernagar, Airport Road, Shahibaug and parts of Chandkheda need airspace permission before anything leaves the ground, and the permitted ceiling in that band is lower than the 120 metres available in a green zone. We check the zone classification of your plot before we quote, because it changes the flight plan, the achievable GSD and the lead time.

Beyond Ahmedabad we work regularly across Gujarat: Gandhinagar and GIFT City, the Dholera Special Investment Region, Vadodara, Surat, Rajkot, Bhavnagar, Jamnagar, Morbi, Mehsana, Ankleshwar, Vapi, Bharuch, and the Kutch belt around Bhuj, Mundra, Gandhidham and Anjar. Kutch sites sit in UTM zone 42N while most of the rest of the state falls in 43N, which is the kind of thing that quietly breaks a deliverable if it is not settled before the first flight.

Where we work across India

Geocartis operates pan-India. Mobilisation is by road for sites within reach of Ahmedabad and by air for everything else, with equipment and control gear travelling with the crew.

Western India. Mumbai, Navi Mumbai, Thane, Pune, Nashik, Chhatrapati Sambhajinagar, Nagpur, Solapur and Goa. Most of our Maharashtra work is industrial land, township development and quarry compliance.

Central India. Indore, Bhopal, Jabalpur, Raipur, Bilaspur, Korba and Singrauli. This is heavy mining and thermal power country, and it is where the annual compliance survey cycle drives most of the demand.

Eastern India. Kolkata, Bhubaneswar, Cuttack, Rourkela, Angul, Keonjhar, Jamshedpur, Ranchi, Dhanbad, Bokaro and Guwahati. Iron ore, coal, steel plants and the port and rail infrastructure that feeds them.

Northern India. Delhi NCR including Gurugram, Noida, Faridabad and Ghaziabad, plus Jaipur, Jodhpur, Udaipur, Chandigarh, Ludhiana, Amritsar, Dehradun, Lucknow, Kanpur, Varanasi, Prayagraj and Patna. Expressway corridors, industrial parks and canal networks dominate here. The airspace over Delhi and along the international border belt carries restrictions that no amount of planning removes, and we say so at quotation stage rather than after mobilisation.

Southern India. Hyderabad, Bengaluru, Chennai, Coimbatore, Madurai, Tiruchirappalli, Visakhapatnam, Vijayawada, Kochi, Thiruvananthapuram and Mangaluru. Solar parks, ports, IT campuses and granite and laterite quarries.

How we do it

The workflow below is the same for a two hectare plot in Bopal and a 900 hectare iron ore lease in Keonjhar. The scale changes, the sequence does not.

Step 1: Scope, coordinate system and accuracy target

Before anything is planned we settle four things with you: the boundary of the survey area, the coordinate reference system, the required ground sample distance, and the list of deliverables.

The coordinate system is the one clients most often leave until later, and it is the one that causes the most rework. Our default for capture is Geographic Coordinate System on WGS84 datum, with projected outputs in UTM WGS84. India spans UTM zones 42N to 47N, so the correct zone depends on where the site is: 42N for Kutch and western Rajasthan, 43N for Ahmedabad, Mumbai and Delhi, 44N for Chennai and Hyderabad, 45N for Kolkata, 46N and 47N for the north-east. If your engineering team works on a local grid, a project datum or a Survey of India sheet, we need the transformation parameters before we set out ground control, not after the point cloud exists.

The accuracy target follows from what you will do with the data. Volume reconciliation on a stockyard, contour generation for a drainage design and as-built comparison against a BIM model are three different jobs with three different tolerances, and pricing a site to the tightest of them when you only need the loosest is money wasted.

Step 2: Airspace check and DGCA permissions

Every flight in India runs under the Drone Rules 2021, administered by DGCA through the Digital Sky platform. Before we quote we check the site against the published airspace map.

Green zones allow operation up to 120 metres above ground level without prior flight permission. Yellow zones, which extend from 8 to 12 kilometres out from an airport perimeter and cover various other controlled areas, require permission from the relevant authority and carry a lower ceiling. Red zones are closed unless the Central Government specifically authorises the flight, which for commercial survey work means the answer is usually no.

Our drones are registered on Digital Sky with active Unique Identification Numbers, and our pilots hold DGCA Remote Pilot Certificates. You are entitled to ask for both, and for any client operating in a regulated sector we supply the UIN, the pilot certificate number and the flight log as part of the handover pack. On mining work these are mandatory fields in the IBM submission annexure, so they need to be correct.

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 the permission of the person the data pertains to. We treat that as the floor, not the ceiling. Raw imagery and processed outputs stay with you.

Step 3: Ground control

Ground control is the difference between a pretty picture and a survey. It is also the part that clients tend to under-budget.

We establish ground control points across the site with a calibrated DGPS instrument, marked as 50 by 50 centimetre high-contrast crosses so they are unambiguously identifiable in the imagery. Density depends on the job, but for mining leases the Indian Bureau of Mines sets the floor: at least five GCPs per square kilometre of lease area or part thereof, a minimum of four where the area is under one square kilometre, and at least three permanent GCPs placed in undisturbed locations so subsequent years can be tied to the same frame. The IBM worked example is a 478 hectare lease requiring 24 GCPs including the three permanent ones. Individual GCP error must come in under 5 centimetres. Lease boundary pillars do not count as permanent ground control.

Separately from the GCPs, we set independent checkpoints. These are surveyed to the same standard but deliberately withheld from the bundle adjustment. They are how we measure accuracy rather than assert it. A survey validated only against the points used to constrain it tells you nothing.

Where a CORS reference station is available we run RTK, and where it is not we run PPK against a base logged over the site. On repeat surveys the permanent GCPs get re-occupied every cycle, which is what allows year-on-year differencing to mean something.

Step 4: Flight planning and ground sample distance

Ground sample distance is set by three numbers: the pixel pitch of the sensor, the focal length of the lens and the height above ground. The relationship is straightforward. GSD equals pixel pitch multiplied by flying height, divided by focal length.

A worked example makes the practical limits clear. Take a 20 megapixel one-inch sensor, 5472 pixels across a 13.2 millimetre width, giving a pixel pitch of about 2.41 micrometres, paired with an 8.8 millimetre lens. Flown at the 120 metre green zone ceiling, that produces a GSD of roughly 3.3 centimetres. Comfortably inside a 5 centimetre requirement, with margin for terrain variation across the site.

The same arithmetic tells you what happens in a yellow zone. Drop the ceiling and the GSD improves, but each photograph covers less ground, so image count and flight time climb. On a large site inside a restricted band, that can turn a one-day job into a three-day job. This is why the airspace check happens before the quotation and not after.

Overlap settings follow the deliverable. For height model generation we plan a minimum of 80 percent forward overlap and 70 percent lateral overlap, which is also what the IBM SOP specifies for mining surveys. Camera angle is nadir, pointed vertically downwards at 90 degrees, for topographic work. Where facades or vertical structures need to be reconstructed we add an oblique pass or a cross-grid, because a nadir-only block reconstructs vertical surfaces poorly.

On sites with significant relief, a quarry with a 60 metre highwall for instance, a constant-altitude flight plan gives you 3 centimetre GSD at the crest and 6 centimetre at the toe. We use terrain-following planning against an existing elevation model so the GSD stays inside specification across the whole block.

Step 5: Capture

Flights are scheduled for the middle of the day when the sun is high, because long shadows fill pits and building footprints with unmatched dark pixels and degrade the point cloud exactly where you need it. We avoid partly cloudy conditions, which produce moving illumination across a block and visible radiometric banding in the mosaic, and we avoid high wind, which costs endurance and image sharpness.

Corridor jobs, meaning transmission lines, canals, highways and railway alignments, get a different plan shape: a narrow multi-strip block along the alignment with additional cross-strips at intervals to control the accumulation of error along the corridor. A single-strip corridor survey without cross-strips will drift, and the drift is not visible in the deliverable until somebody checks it against a control point.

The crew maintains a flight log through the day covering flight numbers, altitudes, times, battery cycles, GCP occupation and any anomaly worth recording. That log becomes part of the handover.

Step 6: Processing

Processing runs in four stages.

Aerial triangulation comes first. The software extracts tie points across overlapping images, then a bundle adjustment solves simultaneously for camera positions, orientations and interior calibration, constrained by the surveyed GCPs. Camera self-calibration matters here, particularly for lens distortion, because an uncorrected distortion model produces the doming effect that makes a flat site look like a shallow bowl.

Dense matching follows, generating the 3D point cloud. Density is a setting, and it is a trade against processing time and file size. We agree it against the deliverable rather than defaulting to maximum.

Classification separates ground returns from everything else: vegetation, buildings, vehicles, plant, spoil. This is the step that produces the difference between a digital surface model and a digital terrain model, and it is the step where automated results need a human to check them. Automatic classification handles open ground well and struggles with dense scrub, terraced slopes and sites where stockpiles look geometrically like natural landform.

Surface and orthophoto generation comes last. The DSM and DTM are gridded to the agreed resolution, the imagery is orthorectified against the surface model to remove relief displacement and camera tilt, and the individual frames are mosaicked with seamlines routed to avoid cutting through buildings and other tall features. Contours are then generated from the DTM at the interval you specify, and smoothed only to the extent that does not move them.

Step 7: Quality control

We compute residuals at the independent checkpoints and report root mean square error separately in horizontal and vertical. The RMSE report states the number of GCPs used, the number of checkpoints held back, the individual residuals, the aggregate figures, the software used and the processing parameters.

We report the measured number. We do not promise an accuracy figure in advance and then produce a report that happens to match it. If a block comes in outside specification, the honest options are reprocessing, additional control or a reflight, and we will tell you which one it needs.

Step 8: Delivery

Deliverables go out in a structured folder set with projection metadata embedded in every spatial file. Nothing ships as an untagged raster. Where the client needs data loaded into an existing GIS or CAD environment, we match the layer naming and attribute schema you already use rather than imposing ours.

Accuracy: what 5 centimetre GSD actually means

Ground sample distance and accuracy are two different things, and conflating them is the most common source of disappointment in drone survey procurement. GSD is how much ground one pixel covers. Accuracy is how close a measured coordinate is to the truth. A survey can have excellent GSD and poor accuracy if the ground control is thin, badly distributed or wrongly surveyed.

What GSD does set is the resolution floor. You cannot reliably identify or delineate a feature much smaller than a few pixels, so at 5 centimetre GSD a painted road marking is visible, a manhole rim is identifiable and a survey nail is not.

As a working rule for a well-controlled photogrammetric block, horizontal RMSE lands in the region of one to two times the GSD, and vertical RMSE in the region of two to three times. At 5 centimetre GSD that suggests horizontal accuracy of roughly 5 to 10 centimetres and vertical of roughly 10 to 15 centimetres. Those are expectations, not guarantees, and the actual figures come from the checkpoint residuals on your specific block.

Four things degrade the result regardless of GSD. Poor GCP distribution, particularly control clustered in the centre with none near the edges, produces error that grows towards the boundary. Dense vegetation prevents the ground being seen at all, so the DTM under a mature canopy is interpolation rather than measurement, and we mark those areas as such. Water surfaces do not generate reliable tie points and appear as noise or holes. Featureless surfaces such as fresh concrete, dry sand flats and fly ash ponds give the matching algorithm nothing to work with, which is why we sometimes place additional temporary targets across large blank areas.

Where the specification demands better than photogrammetry can deliver, or where the ground is under canopy, LiDAR is the right sensor and we will say so.

Deliverables and output file formats

Every job is quoted against a named list of outputs. The table below covers what we produce and the formats they ship in.

Raw imagery is available on request and is retained for the period agreed in the contract. For mining clients the retention requirement is five years, which we build into the data handling plan by default.

Survey types we cover

The technique is common to all of these. What changes is the flight geometry, the control layout, the classification approach and the shape of the output.

Topographic and land survey. Contours, spot levels, boundary features and existing site conditions for design work. The bread and butter of the service, and the largest share of what we fly around Ahmedabad, Gandhinagar and Vadodara.

Mining survey. Lease mapping, pit and bench geometry, overburden dumps, mineral stockyards, volumetric reconciliation and statutory submissions. Concentrated in Odisha, Chhattisgarh, Jharkhand, Rajasthan, Karnataka and the limestone belts of Gujarat and Madhya Pradesh.

Road and highway survey. Alignment corridors, existing carriageway condition, cross sections at chainage, right of way encroachment and earthwork quantities.

Railway survey. Track corridor mapping, formation levels, overhead line clearances, level crossing geometry and adjoining land acquisition mapping.

Canal and irrigation survey. Canal alignment, bank profiles, cross sections, siltation assessment and command area mapping. Substantial demand across the Narmada canal network in Gujarat and irrigation systems in Rajasthan and Madhya Pradesh.

River and floodplain survey. Channel morphology, bank erosion monitoring, floodplain terrain models and flood inundation visualisation built on the DTM.

City and urban survey. Municipal base mapping, property tax survey support, slum redevelopment mapping, road inventory and utility corridor mapping.

Solid waste survey. Landfill volume, legacy waste quantification for biomining tenders, slope stability geometry and post-remediation verification. Municipal corporations across Gujarat, Maharashtra and Rajasthan have active programmes here.

Property and plot survey. Plot boundary verification against records, built-up area measurement, encroachment identification and site documentation for due diligence.

Solar and wind asset survey. Pre-construction terrain models, module row layout verification, thermal inspection of installed arrays and access road mapping. Relevant across the Kutch and Banaskantha renewable belt, the Rajasthan solar parks and the wind corridors of Tamil Nadu and Karnataka.

Transmission line survey. Corridor mapping, tower position verification, conductor sag and clearance assessment, and right of way vegetation encroachment.

Annual compliance surveys

A large share of drone survey work in India is not discretionary. It is a yearly obligation with a fixed window and a fixed submission format, and missing it creates a regulatory problem rather than a project delay. This is the part of our business built around repeat engagement.

Mining: Rule 34A of the MCDR

The Mineral Conservation and Development Rules 2017, as amended in 2021, insert Rule 34A. Any lessee with an annual excavation plan of one million tonnes or more, or a leased area of 50 hectares or more, must carry out a drone survey of the lease plus 100 metres beyond the boundary during April or May each year, and submit the processed digital elevation model and orthomosaic to the Controller General, Indian Bureau of Mines, on or before 1 July. Lessees below those thresholds submit high resolution georeferenced orthorectified satellite imagery instead.

Two further triggers sit alongside the annual cycle. Any lessee submitting a mining plan document or a modification must have carried out a drone survey within the preceding six months. Preferred bidders issued a letter of intent must survey the auctioned block before submitting their mining plan.

The IBM Standard Operating Procedure sets the technical specification, and it is unusually prescriptive. The camera must be a minimum of 20 megapixels RGB, capable of capturing at less than 5 centimetres GSD. Forward overlap minimum 80 percent, lateral overlap minimum 70 percent, camera nadir at 90 degrees. Capture in Geographic Coordinate System on WGS84; orthomosaic delivered in UTM WGS84 in metres. Orthomosaic resolution 5 centimetres or better as GeoTIFF; DSM and DTM at 15 centimetres or better as GeoTIFF. GCP and boundary pillar coordinates in MS Excel in both geographic and projected form, with a point-geometry shapefile plan. An RMSE report. A single polygon shapefile carrying the full land use classification: lease boundary, actual excavation of mineral and waste, mineral storage, sub-grade stack, overburden dump, afforestation, backfilled and reclaimed area, topsoil stack, infrastructure, roads, railways, tailing pond, effluent treatment plant, mineral separation plant, township and green belt. Plus the flight path plan and the survey log sheet.

The submission annexure asks for the drone agency name, the remote pilot licence number, the drone UIN, the aircraft category and type, sensor specification, flight height, GCP count, the DGPS device used, total RMSE, the processing agency, the software used and the UTM zone. Every one of those fields comes from the survey, which means the survey has to be documented properly while it is happening rather than reconstructed afterwards.

We deliver against that specification as a package, not as a set of files you then have to assemble. The three permanent GCPs stay in the ground between cycles, so year two and year three tie directly to year one and change detection is defensible.

Thermal power stations and coal stockyards

Coal-fired generating stations run continuous physical reconciliation between coal received, coal consumed and coal in stock, and the stock figure is the one that has to be measured rather than inferred from a meter. Manual stockpile measurement on a large yard is slow, hazardous on unstable material, and imprecise on the irregular shapes that real stockpiles take.

Drone volumetrics fix the measurement problem, and the periodic capture also produces the ash pond geometry, dyke condition and conveyor corridor imagery that plant engineering needs anyway. Sites across Korba, Singrauli, Angul, Talcher, Chandrapur and the coastal stations of Gujarat and Andhra Pradesh run on this pattern.

Volume figures depend heavily on the base surface used. We state the method explicitly in every volume report: whether the base is a surveyed hard standing, a triangulated boundary plane or a previous epoch surface. Two providers can produce different volumes from identical point clouds simply by choosing different base surfaces, and the number is meaningless without that declaration.

Transmission and distribution corridors

Line corridors need periodic assessment for vegetation encroachment into the clearance envelope, conductor sag, tower verticality and right of way condition. Walking a corridor is slow and much of it crosses ground with no road access. Flying it is faster, and the resulting point cloud supports clearance measurement directly rather than by inspection and judgement.

Construction progress monitoring and photogrammetric marking

On an active construction site the value of a drone survey compounds with repetition. A single flight is a snapshot. Twelve monthly flights are a record of what was built, when, and how it compares to what was drawn.

We fly at an agreed interval and register every epoch to the same control, which allows three things a one-off survey cannot give you. Earthwork quantities between any two dates, computed from the surfaces rather than estimated from cross sections. As-built comparison against the design model, showing deviation as a coloured surface rather than as a list of discrepancies. And a stacked visual record of the site through time that is genuinely useful in a dispute about when something happened.

Photogrammetric marking on buildings under construction is a related service: precise identification and annotation of built features, grid line positions, column locations, floor edges and opening positions, drawn as vector layers against the orthomosaic and the point cloud. The output goes to the site team as DXF, which is what they can actually work with.

When a drone survey is the wrong tool

We would rather tell you this before you engage us than after.

Photogrammetry cannot see through a canopy. On a site with mature tree cover or dense scrub, the ground surface under that cover is interpolated, and if the design depends on those levels you need LiDAR or a ground crew.

Underground and indoor work is out of scope. So is anything requiring sub-centimetre setting out, where a total station remains the correct instrument.

Cadastral boundary determination is a legal process rather than a measurement problem. We can map what is physically on the ground with high precision, and we can overlay it against a cadastral sheet, but the legally operative boundary comes from the revenue authority and the district inspector of land records.

Red zone sites cannot be flown. Defence installations, certain port and refinery perimeters and parts of the border belt sit outside what any commercial operator can legally do.

Monsoon season is a scheduling constraint across most of India, and in Gujarat it removes roughly June to September from reliable flying. Compliance surveys that must be captured in April or May exist partly for this reason. If you leave a statutory survey until the rain arrives, the option has closed.

Working with Geocartis

To quote a job accurately we need five things: the site boundary as a KML or shapefile, the approximate area in hectares, the coordinate system your outputs must land in, the list of deliverables, and your deadline. If you have a previous survey of the same site, that helps us plan control so the two datasets can be compared.

From there we check the airspace classification, calculate the flight plan and image count, and come back with a fixed scope, a delivery date and a price. Where the site sits in a yellow zone or needs permissions beyond the routine, the lead time for those permissions is stated separately so it is visible rather than buried in the schedule.

Site access, safety induction requirements and any restriction on flying times are worth raising early. On live industrial sites they frequently determine how many days the capture takes.

Frequently asked questions

How accurate is a drone land survey?

With well-distributed ground control surveyed by DGPS, expect horizontal RMSE in the range of one to two times the ground sample distance and vertical RMSE in the range of two to three times. At 5 centimetre GSD that is roughly 5 to 10 centimetres horizontally and 10 to 15 centimetres vertically. We measure this against independent checkpoints on your block and report the actual figures.

What is GSD and why does 5 centimetres matter?

Ground sample distance is the ground area covered by one pixel. At 5 centimetres GSD each pixel represents a 5 by 5 centimetre patch of ground. It matters because it sets what is identifiable in the imagery, and because the Indian Bureau of Mines SOP requires 5 centimetres or better for mining compliance surveys.

How large an area can you cover in a day?

It depends on the required GSD, the terrain and the airspace ceiling. As a rough guide, a flat open block at 5 centimetre GSD in a green zone is a single-day capture up to a few hundred hectares. Corridor work, restricted airspace and undulating terrain all reduce that.

Do you handle the DGCA permissions?

Yes. We check the airspace classification during quotation and obtain the flight permissions required for the site. Red zone sites cannot be flown regardless of who applies.

What file formats do you deliver?

Orthomosaic as TIFF and JPEG. DTM and DSM as TIFF. Contours as DXF and SHP. Point cloud as LAS and LAZ. Object marking as DXF and TIFF. CAD basemap as DXF and DWG. Volume reports as PDF and XLSX. All spatial files carry embedded projection metadata.

Do you work outside Gujarat?

Yes. We are based in Ahmedabad and operate across India, including Mumbai, Pune, Nagpur, Indore, Bhopal, Raipur, Delhi NCR, Jaipur, Lucknow, Kolkata, Bhubaneswar, Ranchi, Hyderabad, Bengaluru, Chennai and Visakhapatnam.

Is a drone survey legally acceptable for statutory submission?

For mining leases under Rule 34A of the MCDR 2017, drone survey is the prescribed method rather than an alternative to one. For other regulatory contexts the acceptability depends on the specific authority and the specification they publish, and we will tell you plainly if the deliverable will not satisfy the requirement.

Who owns the data?

You do. Rule 17 of the UAS Rules 2021 requires the operator not to share survey data with third parties without the permission of the person it pertains to, and our contracts reflect that.

Can you survey the same site every year for comparison?

That is the point of the permanent ground control points. Once they are established and surveyed, every subsequent flight registers to the same frame, and differences between epochs reflect real change on the ground rather than differences in how the two surveys were controlled.

Do you sell drones or software?

No. Geocartis is a services company. We fly, process and deliver data.

Get a quote for your site

Send us the site boundary as a KML or shapefile with your required deliverables and deadline, and we will come back with a flight plan, a delivery schedule and a fixed price. If you are working to a Rule 34A deadline, the capture window is April to May and submission closes on 1 July, so the useful time to start the conversation is well before April.