Upload a survey no workstation could open, and get survey-grade bare-earth DEMs, surface and canopy models, and classified point clouds back fast — one automated cloud run, reproducible to the byte, at a quality that meets and exceeds the federal standard (USGS 3DEP).
If you process LiDAR for a living, none of this is news — it's the standard toolchain, and everyone fights it. If these sound like your week, you're who we built this for.
The desktop pipeline is the bottleneck — not flying. Field crews collect faster than legacy tools can process.
The capable tools are expensive and license-locked — seats, dongles, maintenance, and one machine that can run the job.
Conventional DEMs round off the banks, scarps and breaklines your clients are actually paying you to capture.
Legacy desktop pipelines won't reproduce their own surface — rerun the same data and get a slightly different DEM, which is hard to defend in review.
In the desktop stack, classification is a separate manual step — another tool, another license, another day of cleanup.
Four steps from raw flight data to client-ready files — all in the browser.
Send your LAS/LAZ securely to the cloud. No conversion, no preprocessing on your side.
Automated processing produces the bare-earth DEM, DSM, canopy height model at 25 cm, a classified point cloud, and per-cell uncertainty.
Review everything in the browser — 2D map, 3D point cloud, elevation profiles, click-to-identify on any surface, and difference layers against USGS 3DEP or a prior survey. Clouds flown without a camera can be shown in true colour from public imagery.
Upload survey control points for an accuracy report following ASPRS Positional Accuracy Standards, cross-check independently against USGS 3DEP, and export CAD-ready DXF / LandXML / GeoTIFF with FGDC metadata. Aligning to a prior survey? Change detection runs automatically — erosion and deposition volumes, canopy damage and growth — with the maps and statistics as pages in the report.
One upload returns a complete, ready-to-deliver package — terrain and surface models, a classified point cloud, per-cell uncertainty, an accuracy report, CAD/survey files, metadata and change products. The images below are rendered directly from real PRISM·lidarcloud output rasters — coastal mangrove and creek-cut forest sites, real data.
Bare earth recovered beneath heavy coastal canopy — with the subtle relief intact, not flattened into a guess.
The same run that strips canopy from your DEM hands it back as a 25 cm canopy height model, ready for habitat, forestry, and carbon work.
The terrain model with vegetation, buildings and noise removed — banks, scarps and grade breaks preserved, not rounded off.
GeoTIFF · 50 cmThe top-of-canopy / top-of-structure surface — everything the sensor saw, gridded to a clean raster.
GeoTIFF · 25 cmVegetation height above bare earth — ready for habitat, forestry, biomass and carbon work, from the same run.
GeoTIFF · 25 cmEvery point labelled — ground, low / medium / high vegetation, building, water, road and power line — in the same run.
LAS / LAZA raster co-registered to the DEM that says, cell by cell, how well-supported the surface is — where data is dense and where canopy forced interpolation.
GeoTIFF · per cellA client-ready document with ASPRS-convention statistics, datum lineage, alignment residuals and a plain-language interpretation.
Word (DOCX)Standards-compliant geospatial metadata accompanying the rasters — provenance, datum and lineage a reviewer can cite.
FGDC CSDGM · XMLBreakline-ready 3D polylines, points and TIN surfaces in Florida State Plane (ftUS) with NAVD88 heights — rebuilds correctly in Civil 3D-class software.
DXF · LandXMLAn independent cross-check against the USGS national elevation model — signed difference layer and agreement statistics, labelled screening-grade. We also fetch the 3DEP point cloud and run it through the same engine, so the comparison is apples-to-apples — and you can spin the 3DEP cloud next to your own in the 3D browser.
GeoTIFF + report page · 3DEP cloud overlayThe processed area-of-interest footprint as ready-to-share vector boundaries for GIS and your client's records.
KML · ShapefileAlign to USGS 3DEP — the default national reference — or a prior survey of your own, and the engine quantifies what moved: erosion / deposition volumes with a printed minimum level of detection, plus canopy damage and growth, as map and statistics report pages.
3DEP default · or your prior surveyFlown without a camera? Drape the cloud in natural colour from near-flight-time NAIP aerial (US) or Sentinel-2 imagery — a visual layer only, decoupled from the DEM and classification. The source aerial is included and viewable as a dated raster layer in the 2D map, so you always know exactly what — and when — you're looking at.
colorized LAS / LAZ · dated source imagery
One upload, one click, the complete package. The rasters, the classified cloud, the uncertainty, the report, the CAD/survey files, the metadata and the change products — all in one product zip, with no extra steps.
Screenshots of the PRISM·lidarcloud workspace on the sites shown above. Every layer, panel and tool is on tap — during and after processing — when you want to dig in; or ignore them entirely and take the one-click result.
Upload as many sites as you like, then track them all from your account — no emailing yourself job IDs, no guessing whether a run finished. Each job shows exactly where it is and what it will cost, on a page built for the desk and the field alike.
Polished enough to deliver. Rigorous enough to defend.
Banks, scarps and grade breaks are preserved, not smoothed away — measured: scarp crest preserved to ~1 cm, where the reference workflow rounds the same crest off by ~26 cm.
The same input produces the byte-identical deliverable, run after run. When it's questioned in audit, you can prove it.
Ground, vegetation tiers, buildings and water — labeled in the same run.
U.S. outputs default to NAVD88 orthometric heights (rigorous per-cell GEOID18) on State Plane coordinates — county-correct Florida zones — with WGS84 ellipsoidal optional and EGM2008 fallback outside CONUS.
Accuracy statements following ASPRS Positional Accuracy Standards (2014 & 2023 conventions) from your checkpoints, plus an independent cross-validation against USGS 3DEP — in a client-ready report with a plain-language interpretation.
Nothing to install, nothing to maintain, no seats to count. Open the browser, upload, and pay for what you process.
Access the full power through the browser GUI and / or the API, whichever fits your workflow. Drive the whole pipeline from code — a REST API and a one-line Python / CLI client (pip install prism-lidarcloud) upload, poll, and download the product zip. Wire LiDAR into your own automation; the desktop incumbents can't.
Scanned without a camera? PRISM drapes your cloud in true colour from near-flight-time public imagery for a natural 3D view — a visual layer only, so it never touches the DEM or the classification you sign off on.
Send a client the interactive result, not just a zip. One click turns any finished job into a shareable link — they open the maps, 3D cloud and profiles in the browser with no account and no sign-in. You choose the tier: view-only (let them review before the products are signed off) or view + download (also release the full deliverable package). Your account, other jobs and settings always stay private; links expire automatically.
Every deliverable PRISM·lidarcloud ships is backed by a validation chain: cross-validation against the industry-reference workflow, optional control-point checks, independent national references, a public benchmark, and bit-level reproducibility.
The engine is a proprietary, self-contained scientific pipeline — no third-party GIS licenses anywhere in the chain. Each stage is versioned, and every run records the exact configuration that produced it.
Before any classification, the cloud is profiled — point density, canopy structure, ground visibility, terrain roughness. Processing parameters adapt to the regime (coastal mangrove behaves nothing like an open ag field), including automatic adaptation across two orders of magnitude in point density, from sparse aerial collections to dense drone surveys.
Bare earth is separated from vegetation, buildings, and noise — engineered to keep the terrain features generic tools smooth away: bank tops, erosion scarps, ditch inverts, grade breaks. Legitimate sparse ground returns under canopy are preserved.
Bare-earth, surface, and canopy-height models at 25 cm. The ground surface preserves crest lines and scarp lips that conventional interpolation rounds off — so banks and grade breaks survive into the DEM. Every DEM ships with a per-cell uncertainty raster, so you can see exactly where the model is well-supported and where canopy occlusion forced interpolation.
The delivered cloud carries eight ASPRS classes — ground, low/medium/high vegetation, buildings, water, roads, and power lines — tuned to avoid the classic false positives (mangrove canopy is not a building; a dry field is not a lake).
Clouds flown without a camera can be draped in natural colour from near-flight-time public satellite imagery (higher-resolution aerial over the US where available), date-matched to the survey and cloud-masked. It is a visual layer only — strictly decoupled from the engine, so it never changes the DEM, the surfaces, or the classification you sign off on, and every colour is a measured imagery pixel (nothing is invented).
Your surveyed checkpoints produce an accuracy assessment following ASPRS Positional Accuracy Standards (both the 2014 edition and the 2023 second edition conventions). An independent cross-check against the USGS 3DEP national elevation model runs automatically, and you can render signed difference layers against 3DEP or a prior dataset in the browser. Deliverables export with FGDC metadata, and the accuracy report is a client-ready Word document with an AI-assisted, numbers-grounded plain-language interpretation. Processing against a reference adds double-round co-registration on outlier-excluded stable ground and two-epoch change detection — erosion/deposition volumes with a printed minimum level of detection, canopy damage and growth — rendered as report pages. Comparisons against the coarse national raster are explicitly labelled screening-grade in the report.
One upload starts the whole chain. The only decision you make is which reference your survey is aligned to and compared against — a prior survey you upload, the USGS 3DEP national model fetched automatically, or none. Everything else runs without you.
Co-registration runs before differencing, so what the change maps show is change — not misregistration; the alignment residual is printed next to the result. Validation runs against references you don't control: checkpoints you surveyed and a national model we can't tune. And the report discloses everything that was applied — every offset, every comparison, every skipped step — so a reviewer can retrace the run.
The manual tools are for verification and refinement; the pipeline has already done these steps automatically. Upload control points, run the independent 3DEP cross-check, re-run the alignment or apply a vertical adjustment whenever you want to see the evidence yourself — the deliverables never depend on you doing any of it.
Calibration runs on full-scale sites spanning the toughest cover types — coastal mangrove, dense creek-cut forest, and agricultural research plots, 66 to 342 million points each — processed end-to-end and compared against the industry-reference commercial workflow and national reference data.
| What was measured | Result | Reference |
|---|---|---|
| Bare-earth DEM agreement across terrain regimes | within 2–5 cm mean differencemangrove coast, creek-cut forest, agricultural plots | industry-reference commercial workflow |
| Coastal erosion-scarp crest elevation | within ~1 cm of the raw returnsthe reference workflow rounds the same crest off by ~26 cm | raw LiDAR returns |
| Reproducibility of every raster & cloud | byte-identical (MD5-verified)same input ⇒ same output, independent of machine load or core count | repeated end-to-end runs |
| Agreement with the national elevation model | σ ≈ 0.4–0.5 m on dense-forest terrainsurface-to-surface check dominated by the reference's own uncertainty and collection-epoch differences — not PRISM checkpoint accuracy | USGS 3DEP |
| Public benchmark — forest & rural ground filtering | median ≈5% total error (2–11% per sample)terrain-matched configuration, documented; in the range of the best published filters of its class | ISPRS ground-filter test (independent labels) |
| Building removal from the bare-earth DEM | all reference buildings correctly removedroof points classified, ground interpolated from true surrounding terrain | ag research site with known structures |
The engine's development discipline: no change ships unless it is quality-up, or provably neutral — verified on full-scale calibration sites before promotion, every time.
Coastal mangrove, dense creek-cut forest, and agricultural plots — chosen because each terrain breaks a different assumption. Every engine change re-validates against all of them.
Hillshade review from above — and from below. Underside relief exposes classification artifacts that hide in a top-down view. If it isn't clean from both sides, it doesn't ship.
Candidate changes must reproduce the locked production output byte-for-byte wherever they claim no effect. Drift of a single byte fails the gate.
Every DEM ships with a per-cell uncertainty raster showing where the model is well-supported. Supply your own surveyed checkpoints and it is calibrated against them — so "how good is the DEM here?" gets a numeric answer, not just a colour ramp.
Every processed site can be compared against the USGS national elevation model in one click — an outside reference we don't control.
Accuracy statements follow ASPRS Positional Accuracy Standards (Ed. 1 2014 and Ed. 2 2023 conventions), with FGDC metadata on exports. Built to support a licensed surveyor's review — never to replace it.
U.S. outputs default to NAVD88 orthometric heights via rigorous per-cell GEOID18 undulation — not a single site-wide constant — the convention geomatics and commercial-survey clients expect. WGS84 ellipsoidal heights are available as an option, and outside CONUS the engine falls back to the EGM2008 global geoid. Vertical adjustment from your control points is applied with full before/after disclosure in the report.
State Plane coordinate systems with county-correct zone selection (Florida's zone boundaries follow counties, not latitude — we resolve them the way a surveyor would), US survey feet handled exactly.
DXF (3D breakline-ready polylines, points, TIN faces) and LandXML 1.2 surfaces that rebuild correctly in Civil 3D-class software — northing/easting order and all the ingest foot-guns handled.
Every output carries the exact engine and module versions that produced it. Two years from now, you can state precisely how a deliverable was made — and regenerate it.
PRISM·lidarcloud produces deliverables conformant to the USGS Lidar Base Specification (LBS) / 3DEP requirements, and reports vertical accuracy in the ASPRS Positional Accuracy Standards conventions — both the 2014 first edition and the 2023 second edition.
The product set is structured to the USGS 3DEP / Lidar Base Specification requirements — bare-earth DEM, classified cloud, metadata and reporting. 3DEP is a requirements-based program (it publishes specifications, not an algorithm), so we describe the deliverables as conformant to the specification. The 50 cm DEM matches the QL1 grid-cell size.
The report computes ASPRS Positional Accuracy against your surveyed checkpoints — NVA = 1.96 × RMSEz and VVA at the 95th percentile — and states which accuracy class the data supports. We report the class the data earns: the engine meets the 10 cm RMSEz class where the data supports it, demonstrated against ≥20 surveyed checkpoints — never a class asserted without the control to back it.
A calibrated uncertainty raster shipped with every DEM. 3DEP delivers nothing like it — you get a single project-level statement, not a per-cell answer.
A 25 cm DSM and CHM in the same run. 3DEP delivers a bare-earth DEM; the surface and canopy products here have no 3DEP equivalent.
Semantic point-cloud classes plus DXF / LandXML surfaces — neither is a 3DEP deliverable, and both arrive without a separate manual pass.
Honest about the line: we process delivered point clouds — we don't fly the sensor, so we make no acquisition or boresight-calibration claims. We deliver on top of the bare-earth DEM: per-cell uncertainty, surface and canopy models, semantic classification, and CAD/LandXML interchange. A specific accuracy class is only ever stated when it is backed by ≥20 of your surveyed checkpoints.
Most tools stop at a classified cloud. PRISM·lidarcloud carries each job through the steps a survey practice bills time for — CAD interchange, datum bookkeeping, accuracy statistics, and the documentation a reviewing professional expects — in the same one-click run.
DXF and LandXML surfaces in State Plane coordinates with NAVD88 (GEOID18) heights, ready for Civil 3D-class software — scale factor and full datum lineage stamped on every file, so nothing arrives ambiguous.
Accuracy statistics following ASPRS Positional Accuracy Standards, computed against your surveyed control points — plus an independent cross-check against the USGS 3DEP reference surface. Residuals disclosed, not summarized away.
Re-fly a site and the engine co-registers the epochs and quantifies what moved — erosion and deposition volumes with a printed minimum level of detection — turning a repeat flight into a monitoring deliverable.
An AI-assisted report formatted for a licensed professional's review workflow: methodology, datum lineage, residuals, and limitations stated up front, with a plain-language interpretation your client can read.
Where the licensed professional comes in: every PRISM deliverable is clearly stamped DRAFT — not a sealed survey. Outputs are intended for review and adoption by your firm's — or your client's — licensed surveyor or geomatics professional. PRISM does not provide licensed surveying services.
Run your hardest site through it yourself — free trial, full engine, next to your current tool's output and your checkpoints if you have them.
Start your free trial Launch PRISM·lidarcloudUpload and process in the browser — you pay only for the area you run.
SaaS — upload and process in the browser, priced per area.
Run your hardest site through it yourself — free trial, full engine, no card. All it takes to start is your email. Put our deliverables next to your current tools.
Questions before you start? Write to us — happy to help.
Where PRISM·lidarcloud is heading — capabilities in active development, building on the same one-click run you use today.
These are additions to a complete, shipping, validated pipeline — not gaps in it. Everything above already runs in production today; the items below extend it.
Individual-tree segmentation and tree-level 3D structural metrics, derived from the same one-click run that produces your terrain and surface models.
Automated tree-level and stand-level assessment of storm impact and forest-health change between flights — extending the change detection that ships today.
Additional ASPRS feature classes in the delivered point cloud, widening the semantic detail available straight out of the automated run.
The team
Built by the team behind a decade of airborne-LiDAR research from the University of Florida, Florida Agricultural and Mechanical University, and the United States Forest Service.
Associate Professor of forest ecology and geomatics in the School of Forest, Fisheries, and Geomatics Sciences. He co-directs the Spatial Ecology and Conservation (SPEC) Lab and the GatorEye Unmanned Flying Laboratory, the drone-borne LiDAR program at the heart of this work. His research spans spatial scales from the globe to the leaf, carrying him from the Amazon to Hawaii and back to the New England forests he grew up in.
Research faculty in the Center for Latin American Studies and co-director of the Spatial Ecology and Conservation (SPEC) Lab, where she also co-directs the GatorEye Unmanned Flying Laboratory. Her work centers on the human side of landscape change — how people and ecosystems shape each other across the tropics. She is core faculty in UF's Tropical Conservation and Development program.
Biological scientist and Gulf restoration program specialist with the USDA Forest Service, working with federal, state, and private partners to plan and carry out ecosystem and hydrologic restoration across the Gulf Coast. He is co-founder and co-director of FAMU's Center for Spatial Ecology and Restoration, a USDA Forest Service–Florida A&M partnership. His expertise runs through remote sensing, LiDAR, and ecological modeling of southern forests.
GIS program manager for the USDA Forest Service's National Forests in Florida and co-director of FAMU's Center for Spatial Ecology and Restoration. He brings more than two decades of hands-on geospatial experience to applied LiDAR and forest-mapping work across Florida's public lands.
The PRISM app family builds on more than a decade of proprietary GatorEye algorithmic development by Broadbent & Almeyda Zambrano.
The underlying technology was developed at the University of Florida; commercialization proceeds through UF’s technology-licensing process (UF Innovate | Tech Licensing).
Our mission is simple: take the rigorous kind of LiDAR processing that used to demand a specialist, a high-end workstation, and days of work — and put it one click away in the cloud, for anyone who needs to know the exact shape of the earth.