The Mavic 3M Sees More Than Crops: Detecting Buried Structures with Multispectral Drone Imaging

Cropmark analysis is centuries-old archaeology, but a $5,000 agricultural drone now puts it within reach of any farm, project site, or property survey across the Tri-State region.

The Pompeii study you should know about

Last year, researchers at one of the most studied archaeological sites in the world picked up a DJI Mavic 3M — the same multispectral drone we operate at Dragonfly — to map buried structures across the Praedia of Iulia Felix. They weren't using LiDAR. They weren't using ground-penetrating radar. They were using the same NDVI bands a soybean farmer uses to check crop health.

The drone found buried walls, foundations, and unexcavated rooms by detecting how vegetation responded to what was beneath it.

This is not new science. It's just newly affordable.

How vegetation gives away what's hidden

Plants don't grow uniformly. Where ground conditions vary — even by an inch of moisture or a degree of soil temperature — growth varies with it. That variation is invisible to your eye walking across a field but extraordinary in the near-infrared spectrum.

The mechanism splits into two patterns archaeologists have named cropmarks and soilmarks.

Positive cropmarks appear over buried features that hold extra moisture: ancient ditches, filled pits, septic field laterals, old drainage tile still pulling water. Vegetation grows taller, denser, and greener. NDVI lights it up.

Negative cropmarks appear over buried features that block moisture or nutrients: wall foundations, compacted footings, slab edges, buried concrete. Vegetation grows shorter, sparser, and yellows earlier. NDVI shows stress.

The pattern is reliable enough that peer-reviewed archaeological literature has documented the methodology since the 1920s. The Mavic 3M just does it from 200 feet up at 1.2 centimeters per pixel.

What the Mavic 3M actually sees

The Mavic 3M carries four 5MP multispectral sensors plus one 20MP RGB camera. The multispectral bands are Green, Red, Red Edge, and Near Infrared — the same four bands used in every major precision agriculture index.

Three indices we calculate from these bands:

• NDVI (Normalized Difference Vegetation Index): the photosynthesis indicator. Best for cropmarks visible in actively growing vegetation.
• NDRE (Normalized Difference Red Edge): more sensitive than NDVI in mid-to-late season canopy, useful when crops are dense enough that NDVI saturates.
• GNDVI (Green NDVI): more sensitive to canopy stress, useful for catching subtle vigor variations early.

Different indices catch different anomalies. Running all three over the same flight stack triangulates real subsurface features versus weather artifacts.

Real use cases in our region

Pompeii is interesting. Here's what matters for property in Iowa, Nebraska, South Dakota, and Minnesota.

Buried drainage tile mapping. Old field tile installed decades ago is rarely documented on any paper map. When systems fail, finding the lateral lines without trenching the whole field is a needle-in-a-haystack problem. NDVI scans show the moisture differential along functional and failed lines.

Septic field footprint detection. When you buy rural property and need to know where the septic field actually runs — because the survey is 30 years old and the seller doesn't remember — multispectral imaging frequently catches the lateral patterns from above.

Foundation and slab anomaly detection. Old farmsteads often have buried foundations, root cellars, and outbuildings that didn't get demolished, just buried. Before you build, dig a footing, or run new utilities, knowing what's already underground matters.

Archaeological prospection on tribal and historic lands. The Northern Plains have extensive precontact and historic sites that aren't publicly mapped. Multispectral surveys are non-invasive, FAA-compliant, and respectful of cultural protocols when conducted with proper consent and partnership.

Soil compaction anomaly mapping on agricultural land. Compacted strips from old equipment paths and field roads still show up years later through differential plant response.

What this technology cannot do

Honesty matters more than marketing copy. Multispectral cropmark analysis has real limits:

• It does not penetrate the ground. It infers what's below from what's growing above. Bare frozen ground in January shows nothing.
• It requires vegetative or moisture-responsive surfaces. Asphalt parking lots, gravel pads, and recent construction sites are mostly invisible to this method.
• Depth detection is qualitative, not quantitative. We can tell you something is there. We cannot tell you it is four feet down versus eight feet down without ground truth verification.
• It is seasonal. The best cropmark visibility comes during specific moisture and growth stages. A single flight in the wrong week may show nothing while a flight two weeks later shows everything.

For applications requiring precise depth, density, or composition, ground-penetrating radar remains the appropriate tool. Multispectral is the screening tool: fast, cheap, non-invasive, and able to cover hundreds of acres in a single flight.

When to use it

Multispectral subsurface detection makes sense when:

• The area to survey is too large for trenching or GPR sweep economics
• You need fast, non-invasive screening to decide where deeper investigation is justified
• The site has crop, pasture, or vegetation cover that can respond to subsurface variations
• You want documentation suitable for property records, due diligence, or research

It does not make sense when:

• You already know exactly where to look (call a locator service)
• The site is paved, frozen, or recently disturbed
• You need precise depth or material identification

Working with Dragonfly

We are FAA Part 107 certified, $1M insured, and SAM.gov registered for government work. Survey deliverables include the raw multispectral mosaics, calculated index maps (NDVI, NDRE, GNDVI), and an interpretation report flagging detected anomalies with confidence ratings. For sensitive sites, we work under written confidentiality agreements and respect cultural protocols.

Subsurface survey pricing varies by acreage, vegetation cover, and report depth. Contact us for a project consultation.

References

1. Pompeii multispectral UAV study using DJI Mavic 3M for buried structure detection. Published in Virtual Archaeology Review (polipapers.upv.es).
2. Cropmark detection methodology in archaeological remote sensing. Wikipedia: Cropmark.
3. Lasaponara R., Masini N. (2007). Detection of archaeological crop marks by using satellite QuickBird multispectral imagery. Journal of Archaeological Science, 34(2), 214–221.
4. Grassland archaeology using UAV multispectral data. Archaeological Prospection, 2025. Wiley Online Library.