AGENTS.md — Inundații cu probabilitatea de 1%

Despre date / О данных / About the data

RO. Inundații cu probabilitatea de 1%. Conține 19 poligoane extrase din spațiul de lucru risk al serverului geodata.gov.md prin WFS 2.0.0 la data de 2026-08-16. Coordonatele sunt păstrate în sistemul sursă MOLDREF99 / Moldova TM (EPSG:4026).

RU. Затопление с вероятностью 1%. Содержит 19 полигонов, извлечённых из рабочего пространства risk сервера geodata.gov.md через WFS 2.0.0 от 2026-08-16. Координаты сохранены в исходной системе MOLDREF99 / Moldova TM (EPSG:4026).

EN. Flooding, 1% annual probability. Holds 19 polygons extracted from the risk workspace of geodata.gov.md over WFS 2.0.0 on 2026-08-16. Coordinates are kept in the source system, MOLDREF99 / Moldova TM (EPSG:4026).

Overview

Extracted from risk:All_rivers_wd_RP0100 over WFS 2.0.0 on 2026-08-16, matching the server's own count.

Coordinates are MOLDREF99 / Moldova TM (EPSG:4026): projected, metric, unreprojected. ST_Area returns square metres directly.

These are modelled surfaces, not observations. They come from flood modelling under the Apele Moldovei programme. The source publishes no model version, no input data, no date of production, and no accuracy statement, so this catalog can tell you what the polygons say and nothing about how far to trust them. Treat them as indicative rather than regulatory.

Access

Schema

Note Lea_Balea, which carries an underscore where the other names do not. That is the source's spelling and is preserved.

Data quality

• Nineteen rows, one per river system. Each polygon is the whole modelled extent for that system, so a query about a place needs a spatial predicate rather than a name.
• The three return periods are separate collections rather than one collection with a period column. Comparing them means reading all three.
• No model provenance is published. See the catalog's known-issues page.

Example queries

Flood extent by river:

Buildings inside the 1% flood extent, which is the question this collection exists to answer:

Both sides are EPSG:4026, so no transform is needed, but budget for it: the flood polygons are complex, and 5,000 points take about 8 seconds to test, which puts 1.8M buildings near an hour.

Prefilter on the bounding box first. ST_Within against the real polygon is the expensive part, and the bbox column discards most candidates before it runs:

Related collections

risk_all_rivers_wd_rp0200 is the 0.5% probability extent and risk_all_rivers_wd_rp1000 the 0.1%. Rarer floods cover more ground, so the three nest: anything inside the 1% extent is inside the other two.

risk_all_rivers_dam_breach models dam failure rather than rainfall, and is not comparable to the return-period surfaces.

The catalog's AGENTS.md covers the rest.