What a thematic map must show for an EIA

A map inside an Environmental Impact Assessment is not an illustration. It is evidence. It is read by people who will use it to decide whether a project proceeds, and it may be read again years later by someone checking whether what was promised is what was built.

That changes what the map has to do. An attractive map that cannot be interrogated is worth less than a plain one that can.

The map furniture is not decoration

Most maps that come back for revision do not come back because the analysis was wrong. They come back because something structural was missing, and the reviewer could not verify the claim.

A map carrying analytical weight needs all of the following, every time:

• Projection and datum, stated. Without it, no distance or area on the map can be checked.
• Scale, as a scale bar rather than a ratio alone — ratios become meaningless the moment the document is resized or printed differently.
• A legend that matches the map. Classes that appear in the legend but not the map, or the reverse, are read as carelessness and invite closer scrutiny of everything else.
• North arrow.
• Data source and vintage for every layer shown. A land-cover layer from an eight-year-old image is not wrong, but it must be labelled so the reviewer can weigh it.

None of this is difficult. It is simply easy to leave out when a map is produced under time pressure, and it is the first thing an experienced reviewer checks.

One map, one question

The most common substantive problem is a map trying to carry too much. A single sheet showing land use, drainage, sensitive receptors, the project footprint and the study-area boundary communicates none of them well.

The discipline that helps: write down the question before drawing the map. Which drainage lines does the footprint intersect? is a question. Everything about the site is not. If a map supports two questions, it should usually be two maps.

This also makes the map package easier to review, which is not a trivial benefit. A reviewer who can follow the argument moves faster than one who has to reconstruct it.

Classification schemes must be stated, not implied

Land Use / Land Cover mapping is where this matters most. "Agricultural land" is not a self-evident category — it may or may not include fallow, plantation, or seasonally cultivated land, and different schemes draw those boundaries differently.

State the scheme. State the number of classes. If classes were merged from a source dataset, state what was merged into what. A reviewer who can see the decisions can accept them; one who cannot has to assume the worst.

The same applies to accuracy. If a classification was validated, say how and against what. If it was not, say that too. An unvalidated classification is still useful — an unvalidated classification presented as authoritative is a problem.

Change detection needs matched conditions

Comparing land cover between two dates is one of the most persuasive things a map can do, and one of the easiest to do badly.

Two images from different seasons will show different vegetation regardless of whether land use changed. Two images at different resolutions will produce different class boundaries regardless of whether the boundaries moved. Two images processed with different classification schemes are not comparable at all.

Where matched conditions are not available — and often they are not — the honest approach is to say so and describe what the mismatch could do to the result. A stated limitation is a far smaller problem than an unstated one discovered by a reviewer.

Sensitivity mapping earns its value early

Environmental sensitivity mapping is frequently produced late, as documentation of a layout that has already been fixed. At that point it can only describe impacts, not reduce them.

Produced early, the same map changes decisions. A drainage line identified before the layout is frozen can be designed around. The same line identified afterwards becomes a mitigation commitment, an ongoing monitoring obligation, and a permanent constraint on the site.

The map costs the same either way. What differs is what it is still able to influence.

What good looks like

A map package that reviews well tends to share a few characteristics:

1. Each map answers one clearly identifiable question.
2. Every layer is attributed to a source with a date.
3. The classification scheme is written down, not inferred from the legend.
4. Limitations are stated by the author rather than discovered by the reader.
5. Scales are chosen for the decision, and stated.

None of this requires more sophisticated analysis. It requires deciding what the map is for before producing it — which is, in practice, the part most often skipped.

If you are preparing thematic maps for an impact assessment and want a second read before the package goes in, that is work we do regularly — see GIS, remote sensing and mapping and Environmental Impact Assessment and clearance.