We identified and fixed a flood zone accuracy issue on LandLens in April 2026. This post explains what the problem was, what we fixed, and what it means for site assessments run on the platform.
We are publishing this because transparency about data accuracy matters. If you use LandLens to make planning and acquisition decisions, you deserve to know the quality of the data you are working with — including when we find and fix a problem.
What the Issue Was
The Environment Agency's Flood Map for Planning data is delivered via a Web Feature Service (WFS). When LandLens queried that service, it used a bounding box spatial filter to retrieve flood zone polygons near a given postcode.
The problem: the EA's WFS returns any feature whose bounding box overlaps the query area — not features whose actual geometry intersects it. A large flood zone polygon located hundreds of metres away from a postcode can still have a bounding box that overlaps the query window. The previous implementation marked any returned feature as a match, without checking whether the postcode was actually inside the flood zone polygon.
The result was false positives — postcodes being classified as Flood Zone 3 when they were not inside any flood zone at all.
Which Postcodes Were Affected
The issue only produced false positives — postcodes incorrectly flagged as being in a flood zone. It did not produce false negatives. Postcodes genuinely inside a flood zone were correctly identified.
The postcodes most likely to be affected were those located close to — but not inside — a flood zone boundary. Typically these were rural postcodes near rivers or streams where a flood zone polygon existed nearby but did not cover the postcode itself.
Three specific examples identified during testing:
- GU28 9JH (Petworth, West Sussex) — returned as Zone 3, actually Zone 1. Nearest flood zone boundary was 131 metres away
- LA22 9QQ (Ambleside, Cumbria) — returned as Zone 3, actually Zone 1. Nearest boundary 33 metres away
- SY4 1AA (rural Shropshire) — returned as Zone 3, actually Zone 1. Nearest boundary 338 metres away
Urban postcodes directly adjacent to rivers and genuinely inside flood zones were not affected — those were correctly flagged before and after the fix.
What We Fixed
We built a new flood check module that works correctly in three steps:
Step 1 — Tighter pre-filter
The query window was reduced from ±110 metres to ±165 metres, which is still generous enough to catch postcodes near flood zone boundaries while reducing the number of false candidates returned.
Step 2 — Force correct coordinate system
The EA WFS defaults to British National Grid coordinates. The previous implementation did not convert these correctly before comparing against postcode coordinates in WGS84. The new module forces the response geometry into WGS84 explicitly.
Step 3 — True point-in-polygon check
Every flood zone polygon returned by the WFS is now tested geometrically to confirm whether the postcode coordinate actually falls inside it. Only polygons that contain the point are counted as a match.
This fix was applied across all three platform routes that query flood zone data — the constraint map, the bulk postcode search tool, and the site report generator.
Verification
Post-fix testing confirmed correct results across six test cases — three rural postcodes that should return no flood zone classification, and three known flood-prone locations that should return Zone 3 or Zone 2:
| Location | Expected | Result |
|---|---|---|
| GU28 9JH — Petworth | Zone 1 | ✅ Correct |
| LA22 9QQ — Ambleside | Zone 1 | ✅ Correct |
| SY4 1AA — rural Shropshire | Zone 1 | ✅ Correct |
| Thames foreshore, London Bridge | Zone 3 | ✅ Correct |
| Tewkesbury town centre | Zone 3 | ✅ Correct |
| Reading, Thames riverside | Zone 2/3 | ✅ Correct |
What This Means for Previous Site Assessments
If you ran a constraint map search, bulk postcode search or generated a site report before this fix was deployed in April 2026, and the result showed a Flood Zone 3 classification for a postcode that is not directly adjacent to a river or coastal flood area, that result may have been a false positive.
The fix only affected false positives — sites genuinely in Flood Zone 2 or 3 were correctly identified before and after the update. If LandLens told you a site had no flood risk, that result was correct.
Going Forward
Flood zone data on LandLens now uses true geometric point-in-polygon verification against the Environment Agency Flood Map for Planning WFS. Every constraint map search, bulk search and site report generated from April 2026 onwards uses the corrected methodology.
We will continue publishing platform updates like this one when we identify and resolve accuracy issues. Data quality is the foundation of everything LandLens is built to do — and we will always be transparent when we find something that needs fixing.