Key Takeaways:
- Hochwassergefahrenkarten (flood hazard maps) come from the EU Floods Directive (2007/60/EC), implemented via §74 WHG; they model flood scenarios by probability class and are informational, not a legal designation
- Überschwemmungsgebiete (legally designated flood areas) are designated by state ordinance under §76 WHG, and §78 WHG attaches binding building restrictions to that boundary
- A site can sit inside a hazard scenario without lying in a designated flood area, and vice versa; screening for only one of the two misses real risk
- Coverage differs by layer: Hochwassergefahrenflächen cover 13 of 16 German states (Bavaria, Saxony-Anhalt and Thuringia do not publish at source), while Überschwemmungsgebiete cover a different 13 of 16 (Baden-Württemberg, Bavaria and Saxony-Anhalt do not publish at source)
- Used together, the two layers answer the two questions a lender or insurer actually asks: how exposed is this site, and is building here legally restricted
If a due-diligence pack for a German site lists "flood risk: yes," the next question is which kind of yes. Germany produces two distinct categories of flood data, and they get confused constantly because both carry the word "flood" and both show up as polygons on a map. Mapular's Hochwassergefahren layer covers one half of that picture; a separate layer covers the other. This post explains what each one actually claims, why the difference matters to anyone signing off on a site, and how to use both without mixing them up.
What question is a bank or insurer actually screening for?
Ask a lender or an insurer what they need from "flood data" and you get two answers bundled into one request. The first is a risk question: how likely is water to reach this site, and how bad could it get? The second is a legal question: is this site inside a boundary where the law restricts what can be built?
Those are not the same question, and German flood data does not answer them with a single dataset. One category of map models the first question. A different, legally distinct designation answers the second. Treating them as interchangeable is how a screening report ends up either overstating risk on a site that carries no legal restriction, or missing a restriction on a site that never triggered a hazard scenario.
What do Hochwassergefahrenkarten actually show?
Hochwassergefahrenkarten (flood hazard maps) are produced by the German states under the EU Floods Directive (2007/60/EC), implemented domestically via §74 of the Wasserhaushaltsgesetz (WHG, the federal water act). For each probability class, the maps model how far a flood of that magnitude could plausibly reach. The classes are a frequent scenario (high probability, often in the range of HQ10 to HQ30), a medium scenario (typically the 100-year flood, HQ100), and an extreme scenario with low probability (HQextrem).
That modelling is informational by design. The areas trigger no direct building ban. Their weight comes from how authorities, insurers and lenders actually use them: authorities consult the maps in planning, insurers and lenders use the scenario classes to grade exposure instead of applying a flat flood flag, and the hazard areas turn up routinely as findings in due-diligence reports. A site touched only by the extreme scenario is a materially different finding from one sitting inside the frequent scenario, and that distinction is the entire value of having three classes instead of one binary flag.
Mapular's Hochwassergefahrenflächen layer unifies these published hazard areas across 13 of 16 German states, with the scenario class, geometry type and source state carried as attributes on every feature, so a query can distinguish "frequent" from "extreme" without opening a state's map viewer.
What are Überschwemmungsgebiete?
Überschwemmungsgebiete (legally designated flood areas) are a different thing entirely. They are formally designated under §76 WHG: areas that statistically flood at least once in a hundred years, or that are needed for flood retention. Designation happens by state ordinance, and that ordinance makes the boundary legally binding, not modelled.
The consequence is §78 WHG. Inside a festgesetztes (formally designated) Überschwemmungsgebiet, §78 prohibits the designation of new development zones and puts new construction under a strict permit requirement, typically with conditions such as flood-adapted building and compensation for lost retention volume. Areas that are vorläufig gesichert (provisionally secured) ahead of formal designation carry the same requirements accordingly. This is not a risk score. It is a legal act that authorities, courts and permitting procedures work from directly.
Mapular's Überschwemmungsgebiete layer carries the designation status, name, waterbody and source state on every polygon, which is what lets a screening query flag "§78-restricted" as a distinct finding rather than folding it into a general flood-risk score.
Why can a site sit in one and not the other?
Because the two categories are built on different logic, a site can appear in a hazard scenario without lying in a designated flood area, and the reverse holds too. A parcel might sit inside the extreme (HQextrem) hazard scenario, which is informational and imposes nothing directly, while sitting entirely outside any festgesetztes Überschwemmungsgebiet, meaning no §78 restriction applies. Conversely, a parcel can lie inside a designated Überschwemmungsgebiet, with the full weight of §78 WHG attached, even where the hazard-map scenario for that stretch of river was drawn conservatively or generously depending on the state's own modelling conventions.
For a screening workflow, that means checking only the hazard layer can miss a binding legal restriction, and checking only the designated-areas layer can miss a real exposure signal that has not yet been formalised into an ordinance. Neither layer substitutes for the other. A thorough site check runs both.
Why doesn't Germany's flood data cover all 16 states?
Both layers carry the same structural gap for the same reason: publication is a state matter, and not every state publishes.
Hochwassergefahrenflächen are available for 13 of 16 states. Bavaria, Saxony-Anhalt and Thuringia do not publish their flood hazard areas at source. Überschwemmungsgebiete are also available for 13 of 16 states, but not the same 13: Baden-Württemberg, Bavaria and Saxony-Anhalt do not publish their designated flood areas at source. Bavaria and Saxony-Anhalt are missing from both. Thuringia is missing only from the hazard layer. Baden-Württemberg is missing only from the designated-areas layer.
That asymmetry is exactly why the two datasets need to be checked as separate layers rather than assumed to move together. Mapular states each gap openly on the respective layer page rather than filling it with a modelled substitute that would not match the official source the authorities themselves work from. A dataset used for legal or underwriting purposes has to stay traceable to what a state actually published, even when that means saying "not available here."
How do you screen with both layers as one model?
In practice, the two layers answer complementary parts of the same portfolio question. Run the designated-areas layer first to flag any site inside a festgesetztes or vorläufig gesichertes Überschwemmungsgebiet: that result is a binding legal finding, in or out, and it belongs in a due-diligence report as such. Then run the hazard layer across the same portfolio to grade every remaining site by scenario class, frequent, medium or extreme, which turns "is there flood risk?" into a graded answer instead of a yes/no guess.
Combined, a query can answer, per site: is this legally restricted under §78, and separately, how exposed is it under the EU Floods Directive scenarios. That is the shape of the question a lender or insurer is actually asking when they request "flood data," and it is why both layers sit in Mapular's Planning & Building Regulations data, alongside the other flood-relevant reference layers in the geospatial data catalog.
Where to go from here
If your flood screening today means opening one PDF map per state and eyeballing whether a parcel falls inside a shaded zone, the fix is treating both categories as queryable data instead of documents. Start with the Hochwassergefahren layer page: it lists the scenario classes, the per-state coverage, and how to request sample data to check the schema against your own workflow before you talk to anyone about a contract.
