What is an environmental baseline map?

It is a documented snapshot of the existing environment before a project begins, covering physical, biological and human receptors, against which future impacts and monitoring results are compared in an environmental impact assessment.

What layers does an environmental baseline need?

Typically hydrology and watersheds, soils and geology, land cover, protected areas and habitats, surface and groundwater features, air and noise context, cultural heritage, and human receptors such as dwellings and water supplies, each with source and date metadata.

How do I design buffers around sensitive receptors?

Use distance buffers tied to regulatory or guidance thresholds (for example a setback from a watercourse or a protected habitat), computed in a projected CRS so distances are in true metres, and document the threshold and its source.

How do I keep baseline data defensible for an EIA?

Record source, date, accuracy, CRS and licence for every layer; use authoritative datasets where they exist; keep raw and derived data separate; and version the baseline so the snapshot used in assessment is reproducible years later.

GIS for Environmental Baseline Mapping

Short answer

An environmental baseline map is a documented, dated snapshot of the existing environment across a study area, built so that future impacts and monitoring can be measured against it. In GIS this means assembling receptor and constraint layers — hydrology, soils, geology, habitats, protected areas, cultural heritage, and human receptors — with strict CRS discipline, distance buffers tied to real regulatory thresholds, and metadata thorough enough to survive an Environmental Impact Assessment (EIA) review years later.

The defining quality of baseline work is not analytical cleverness; it is provenance and reproducibility. A baseline that cannot prove where each layer came from, when, and at what accuracy is worthless when the project is challenged.

What a baseline must capture

Group the layers by receptor type so nothing is missed:

Physical — geology, soils, terrain (DEM, slope), watersheds, surface water (rivers, lakes, wetlands), groundwater (aquifers, wells, springs), flood extent, coastal/marine where relevant.
Biological — land cover/vegetation, habitats, protected sites (e.g. Natura 2000, Ramsar wetlands, national designations), and species records.
Human — dwellings, settlements, drinking-water sources, agricultural land, recreation, and infrastructure.
Cultural — archaeological and heritage sites, scheduled monuments, landscape designations.

For each, the rule is the same: prefer an authoritative source (national mapping agency, environment agency, Copernicus land monitoring, geological survey) over a digitised approximation, and record its identity.

CRS discipline — buffers are measured in metres, not degrees

The most common and most damaging baseline error is computing distance in a geographic CRS. A buffer of "250" around a watercourse in EPSG:4326 (WGS84 lat/long) is 250 degrees, which is meaningless. Every distance operation must run in a projected CRS whose units are metres — a national grid or the appropriate UTM zone.

Set this up explicitly. In PostGIS:

-- transform once into a metre-based CRS, then buffer
SELECT id, ST_Buffer(ST_Transform(geom, 25831), 250) AS buf250
FROM watercourses;

(EPSG:25831 is ETRS89 / UTM 31N; choose the correct zone for your area.) With ogr2ogr the same discipline applies — reproject before any distance work:

ogr2ogr -t_srs EPSG:25831 watercourses_utm.gpkg watercourses.gpkg

Confirm the units before trusting any buffer: a quick check that a known 1 km feature measures ~1000 in the working CRS catches the error immediately.

Buffers and setbacks tied to thresholds

Buffers in a baseline are not arbitrary — each should encode a regulatory or guidance threshold and cite it. Examples:

A riparian setback from watercourses (the distance comes from the relevant water-protection regulation, not from preference).
A protection buffer around a designated habitat or drinking-water source.
A blast/dust/noise standoff from dwellings drawn from the applicable guidance.

Build them as distinct layers with the threshold value and its source in the attributes, so a reviewer reads not just where the constraint is but why it has that radius. Combine overlapping constraints into a composite "environmental constraints" layer with ST_Union, but keep the individual buffers so the contribution of each is traceable.

Sampling design for new survey data

Where the baseline requires field data (water quality, soil contamination, noise, ecology), GIS designs the sampling. Use a defensible spatial scheme — a regular grid, a stratified-random design weighted by receptor sensitivity, or targeted points at suspected sources and at control/background locations upstream or upwind. Generate the points in GIS, attribute them with the rationale (type = background | source | compliance), and export coordinates for the field team. Recording background/control stations is essential: an impact is only demonstrable against an unaffected reference.

Worked example: a quarry extension baseline

Define the study area and a survey buffer around the proposed extension.
Assemble authoritative layers: geology and soils from the geological survey, hydrology and aquifer designation from the environment agency, land cover from Copernicus CLC, protected sites from the national register, dwellings from address data.
Reproject everything to the national grid (metre units).
Buffer watercourses, the drinking-water source, and the nearest dwellings at their regulatory setbacks; union into a constraints layer.
Identify groundwater dependence: intersect the proposed footprint with the aquifer layer to flag where excavation approaches the water table.
Design a stratified water- and soil-sampling network with background stations upstream.
Freeze the baseline as a versioned snapshot (date-stamped GeoPackage plus a metadata record).

The product tells the assessor exactly which receptors lie within influence distance and where monitoring must establish pre-project conditions.

Metadata that survives review

Each layer needs, at minimum: source organisation, dataset name and date, accuracy/scale, CRS, licence, and processing history (what you did to it). Adopt a recognised metadata standard (ISO 19115 / the INSPIRE profile in Europe) where possible, and keep raw and derived data separate so the chain from authoritative source to constraint layer is unbroken. Version the entire baseline: an EIA may be examined years after submission, and "the land cover as it was on 2026-03-07" must be reproducible exactly.

Common pitfalls and why they happen

Buffering in degrees. Distances become nonsensical because the CRS units are angular, not metric. Reproject to metres first and verify.
No background/control samples. Without an unaffected reference, no impact can be quantified against the baseline. Design them in from the start.
Mixing datasets of different vintages silently. A 2010 land cover next to 2025 imagery misrepresents the "current" state; date every layer.
Digitising over authoritative data. A hand-drawn habitat boundary is less defensible than the official designation; use the source of record.
Losing provenance. A layer with no source or date cannot be defended when the assessment is challenged, regardless of how accurate it actually is.

QA and validation

Confirm every layer is in the working metre-based CRS and that a known distance measures correctly; check that each buffer cites its threshold and source; verify completeness of metadata (source, date, accuracy, licence) on all layers; ensure raw and derived data are separated and the snapshot is versioned; and reconcile the constraints layer against the individual buffers so nothing is double-counted or lost in the union.

Bathyl perspective

We treat environmental baselines as evidence, not illustration. Every layer carries its source, date, accuracy, and CRS; distances are computed in true metres against cited thresholds; and the whole baseline is frozen as a reproducible, versioned snapshot. When the assessment is questioned, the map answers for itself.