Australia has one of the world's best free regional LiDAR datasets. ELVIS covers most of the populated coast and increasing inland coverage; state portals fill gaps; many councils publish their own captures. For some projects it's all you need. For others, the differences between regional and hyper-local don't show in headline accuracy but show in everything that depends on it — recency, density under canopy, classification quality, hydro-enforcement, edge handling. The question isn't 'which is better'; it's 'which is appropriate for this specific project'.
If you've ever started a project on free ELVIS data and ended up commissioning a fresh capture after discovering the regional data couldn't support your specific use case, you've met the regional-vs-hyper-local decision point. The regional data was technically correct for what it was. It just wasn't enough for what you needed — either because it was captured five years ago and the site has changed, or because it was classified to generic standards and your project needed specific classes, or because it was processed with no hydro-enforcement and your hydraulic model needs it.
This is the natural follow-up to the ELVIS vs commissioned LiDAR article — which laid out the general comparison. This article goes one level deeper into the decision framework: eight specific project characteristics that make hyper-local commissioned capture the right call, three scenarios where regional data is genuinely fine, and the four-step test that distinguishes them before commissioning.
The decision matters because the cost gap is real. Regional data is free or near-free; hyper-local capture is $5k-50k+ depending on project shape. Defaulting to commissioned capture when regional would have worked wastes budget; defaulting to regional when hyper-local was needed produces deliverables that don't support the design intent.
Free Australian regional LiDAR coverage typically delivers:
Coverage area. ELVIS provides reasonable coverage of populated coastal areas and increasing inland coverage. State portals (Spatial Services NSW, VicMap, QSpatial, etc.) supplement with state-specific captures. Council LGAs increasingly publish their own captures via state or council platforms.
Headline accuracy. Typical regional captures are specified at ±150-300 mm vertical RMSE at LPI / state quality standards. Recent captures (post-2020) often exceed spec at ±100-200 mm.
Density. Typical regional density is 4-8 ppm² ground returns; some recent captures hit 8-15 ppm². The density is engineered for whole-state programmes, not for project-specific deliverables.
Classification. ASPRS standard classes applied via automated processing; ground, vegetation strata (typically merged), buildings, water. No project- specific cleanup, no extended classes.
Capture age. Highly variable by location. Metropolitan areas often have data from the last 3-5 years; rural areas often 5-10 years old; some areas have data from a decade or more ago that's still the most recent available.
Hydro-enforcement. Generally none. State captures prioritise broad terrain accuracy over structure- specific enforcement.
Format. Typically LAS or LAZ point cloud plus DEM raster (usually 1 m grid or coarser). Sometimes DTM at finer resolution where state programmes included civil-grade processing.
For planning-grade work, broad analytics, large-area modelling and many feasibility-stage projects, this is genuinely sufficient.
(See accuracy article for what these numbers actually mean for deliverables.)
The project needs engineering-grade accuracy (±20-30 mm RMSE) or tighter. Regional data at ±150-300 mm doesn't satisfy the spec regardless of how cleanly it's processed. The headline number is the dispositive factor.
Common project types. Engineering DTM for infrastructure design, construction setting-out, asset-condition monitoring, deformation analysis, compliance-grade measurement.
The site has changed materially since the most recent regional capture. New construction, substantial earthworks, completed development, fire or flood event, mining progress.
Common project types. Post-construction as-built, current-condition baseline for new project, post-event damage assessment, mining volumetrics.
The diagnostic. Compare regional capture date against last significant site change. If change post-dates capture, regional data shows the past; project needs the present.
The project needs accurate ground under closed or moderate canopy. Regional captures at 4-8 ppm² total density typically have only 1-2 ppm² ground returns under canopy — too sparse for engineering- grade DTM in vegetated areas.
Common project types. Forestry inventory, wildfire fuel mapping, drainage design through vegetated terrain, environmental impact assessment with vegetation overlay.
The fix. Hyper-local capture at 80-150 ppm² total density produces 20-40 ppm² ground returns under typical canopy — enough for engineering work.
(See point density article for density-by-deliverable specifics.)
The project needs extended ASPRS classes (powerline conductors, transmission towers, road surface, rail infrastructure) or custom user classes. Regional data ships with standard classes only; the specific classes aren't populated.
Common project types. Powerline corridor work, utility asset capture, rail corridor projects, specific infrastructure inventory.
(See quality vs speed classification article for what targeted classification actually delivers.)
The project will use the DTM for drainage, flood modelling or hydraulic design and needs bare-earth enforcement at bridges, culverts, weirs and low- level crossings. Regional data doesn't enforce these structures; the bare-earth DTM carries the bridge deck into the surface where it shouldn't be.
Common project types. Flood modelling, drainage design, stormwater infrastructure, catchment hydraulics.
(See flood modelling article for why this matters specifically.)
The project has a hard deliverable boundary (property line, contract area, specific catchment) and needs the captured area to extend past the deliverable boundary so edge effects fall outside the AOI. Regional data may have edge effects inside the AOI that aren't fixable from the existing data.
Common project types. Property-boundary projects, contract-defined deliverables, infrastructure work with hard boundaries.
(See edge effects article for the buffer-strategy framework.)
The project uses a local site grid, non-standard datum, or specific projection that regional data isn't in. Transforming regional data into the target system adds error and may not be possible without re-processing.
Common project types. Mine site projects, infrastructure-corridor projects with project- specific grids, integration with legacy GDA94 datasets.
(See coordinate transformations article for transformation failure modes.)
The project will use this capture as the baseline for change-detection across multiple future cycles. The baseline needs to be at the same accuracy and processing standard as future captures — typically hyper-local engineering-grade rather than regional planning-grade.
Common project types. Settlement monitoring, asset-condition programmes, vegetation regrowth tracking, mining or quarry progress.
(See change-detection article for the methodology requirements.)
fine
Project is at concept or feasibility stage; the deliverable supports option assessment rather than detailed design; accuracy tolerance is generous (±100 mm vertical is fine); the site is captured and recent in regional data.
Examples. Initial route corridor screening, broad-scale renewable energy site assessment, catchment-scale hydraulic feasibility, planning- permission terrain reference, broad bushfire hazard mapping.
The project area is larger than is economical to capture hyper-local (thousands of hectares to hundreds of square kilometres); the analysis operates at regional scale; the use case tolerates generic processing.
Examples. River basin modelling, regional biodiversity mapping, statewide forest inventory, climate-impact assessment.
The project needs to understand terrain as it was at some past date; regional data from that date is the only available record. Even if accuracy is limited, it's the only data available — hyper-local re-capture today gives you current state, not historical state.
Examples. Post-event change quantification (pre-event from regional, post-event from hyper- local), historical settlement analysis, vegetation change over decades.
A practical test for distinguishing regional-sufficient from hyper-local-needed projects:
1. Check the accuracy spec required vs regional spec. If your project needs ±30 mm or tighter, regional data probably can't satisfy regardless of other factors. Stop here; hyper-local needed.
2. Check the capture age vs site change. If the site has changed materially since the regional capture, the data describes the past. If your project needs current state, hyper-local needed.
3. Check the use-case requirements against regional processing. Does your project need extended classification, hydro-enforcement, edge buffer, custom CRS, or monitoring baseline? Each is a tick toward hyper-local; multiple ticks make hyper-local the right call.
4. Run the cost-benefit comparison. Hyper- local capture cost vs the cost of the project problem regional data would cause. For high-value or compliance-critical projects, hyper-local typically pays back; for low-value or feasibility work, regional is often the better economic answer.
A useful default: planning or feasibility work defaults to regional; engineering or compliance work defaults to hyper-local; mid-tier work needs explicit analysis.
Sometimes the right answer is hybrid: regional data for the broad area, hyper-local for the specific zone where accuracy matters.
Examples.
The hybrid approach works when the project has naturally tiered accuracy needs. It requires careful integration to avoid datum mismatches and seam artefacts (see coordinate transformations article for the failure modes), but for large projects with focused engineering zones, it's often the most cost-efficient approach.
regional data
Three procurement implications:
Lead time. Regional data is available immediately (typically within hours of search and download). Hyper-local capture has weeks to months of lead time including operator scheduling, capture, processing and QA.
Cost. Regional data is free or near-free (some state portals charge nominal fees; high-value commercial datasets cost $0-10k). Hyper-local capture costs $5k-50k+ depending on project shape.
Control. Regional data is what it is — unchanged after the fact, no project-specific adjustments possible. Hyper-local capture can be specified, validated and adjusted to project needs.
Provenance and continuity. Regional data comes with public metadata; long-term continuity is the state's responsibility. Hyper-local capture comes with operator-supplied documentation; archive retention is part of the contract.
(See archive retention from procurement checklist for how to handle the continuity question in hyper-local contracts.)
Three patterns we see when this decision goes badly:
Defaulting to commissioned when regional was fine. A planning-grade feasibility study commissions $25k of hyper-local capture when ELVIS data would have supported the same analysis. Cost saving available, missed.
Defaulting to regional when hyper-local was needed. Engineering design starts on regional data; ±200 mm vertical accuracy turns out to be insufficient for the drainage modelling; project re-baselines on new capture six months in. Schedule slip, sunk-cost design rework.
Not checking regional data for currency. Regional data is downloaded; analysis proceeds; team discovers the data is from 2018 and the site was substantially redeveloped in 2023. Findings revised based on actual current state.
The four-step test prevents all three patterns with maybe 30 minutes of analysis at project inception.
Free Australian regional LiDAR (ELVIS, state portals, council LGAs) delivers ±150-300 mm vertical accuracy at 4-8 ppm² density with ASPRS standard classification and no hydro-enforcement. Sufficient for planning, broad analytics, historical reconstruction.
Eight scenarios where hyper-local wins: high accuracy spec, recent site change, dense canopy ground recovery, project-specific classification, hydro-enforcement at structures, edge buffer requirement, non-standard coordinate system, change-detection baseline.
Three scenarios where regional is genuinely fine: broad planning and feasibility, catchment or regional-scale analytics, historical baseline reconstruction.
Four-step test: accuracy spec, capture age vs site change, use-case requirements, cost-benefit. Two or more "hyper-local needed" answers make hyper-local the right call.
Hybrid approach works when the project has naturally tiered accuracy needs — regional for context, hyper-local for the engineering zone.
Three common decision mistakes: defaulting commissioned when regional fine, defaulting regional when hyper-local needed, not checking regional currency. All preventable with 30 minutes of analysis at project inception.
If you're scoping a project and unsure whether regional data would do the job, the four-step test takes about 30 minutes. Send through the project shape, accuracy requirements, site change history, and use case. We'll walk through whether ELVIS is genuinely fine, where hyper-local is warranted, or whether a hybrid is the right answer.
The broader regional-vs-commissioned comparison this article extends. Useful prerequisite for understanding the general framework before applying the eight-scenario test.
The cost picture that determines the economic side of the regional-vs-hyper-local trade-off. Useful for sizing the cost-benefit analysis in the four-step test.