Moldova Geodata

RO. Catalog cloud-native cu date selectate din geoportalul național geodata.gov.md. Româna este limba de referință. Textele în rusă și engleză sunt traduceri. Catalogul nu este o publicație oficială a AGCC.

RU. Облачный каталог избранных данных национального геопортала geodata.gov.md. Румынский текст является основным; русский и английский тексты — переводы. Каталог не является официальной публикацией AGCC.

EN. A cloud-native catalog of selected data from Moldova’s national geoportal, geodata.gov.md. Romanian is authoritative; Russian and English are translations. This is not an official AGCC publication.

Contents

root

AGENTS.md

Details

Visibility: Public
Owner: Nissim Lebovits
Created: 17 Aug 2026
Last Updated: 17 Aug 2026

Analytics

Contents

Details

Analytics

Moldova Geodata

Contents

root

AGENTS.md

Details

Visibility: Public
Owner: Nissim Lebovits
Created: 17 Aug 2026
Last Updated: 17 Aug 2026

Analytics

28 days

Downloads

9,660

Data served

3.4 GB

Countries

28-day downloads

29 Jul 2026

25 Aug 2026

Analytics

Object Preview

Name: AGENTS.md
Size: 4.54 KB
Content Type: text/markdown; charset=utf-8
Last Modified: 17 Aug 2026
Source URL: https://data.source.coop/nlebovits/moldova-geodata/AGENTS.md
Cloud URI: s3://us-west-2.opendata.source.coop/nlebovits/moldova-geodata/AGENTS.md

Download

Object Preview

Agent guide for the Moldova geodata catalog

Despre catalog / О каталоге / About the catalog

RO. Catalogul conține date geospațiale selectate din geodata.gov.md. Româna este limba de referință. Textele în rusă și engleză sunt traduceri.

RU. Каталог содержит избранные геопространственные данные geodata.gov.md. Румынский текст является основным; русский и английский тексты — переводы.

EN. This catalog contains selected geospatial data from geodata.gov.md. Romanian is authoritative; Russian and English are translations.

Data access

Downloadable collections contain GeoParquet in MOLDREF99 / Moldova TM (EPSG:4026). This CRS is projected and uses metres. PMTiles are Web Mercator. Collections whose names start with wms_ contain service links and previews, not downloadable source data.

DuckDB may interpret EPSG:4326 in latitude-longitude order. Use OGC:CRS84 for longitude-latitude order, or set geometry_always_xy = true.

Spatial filtering

Files are Hilbert-ordered and include a bbox struct with GeoParquet 1.1 covering metadata. DuckDB does not add the bounding-box filter automatically. Apply it before the exact spatial predicate:

The coordinates are metres; this example uses a 10 km square. Measured catalog queries were 3–7 times faster with the explicit bbox filter. Performance varies by collection and query.

Joins

Use cadastru_data_uat2 for aggregation by second-level administrative unit. It contains 35 units and uses the same CRS as the other downloadable collections.

Prefer spatial joins. Fields such as gid and objectid are identifiers within a source layer. The source does not document them as foreign keys between layers.

Base map

maps_base_map_2017 combines source layers that share a schema. Use layer_name for the feature class and source_layer for the exact WFS layer. It has separate point, line, and area assets.

The source has incorrect layer_name values on part of the spot-height layer. Use source_layer when exact source membership matters.

Data limitations

The catalog is curated. Absence from this catalog does not prove absence from the source portal.
Coded columns are preserved without interpretation when the source provides no code list.
Canonical source titles may contain spelling errors. Corrected translations are stored in the language-specific fields.
Canonical language varies: maps uses English and the other workspaces use Romanian. Select a language-specific field when a fixed language is required.
Some source rows have null geometry. Filter with geometry IS NOT NULL before spatial operations.
The GeoServer backends return inconsistent layer metadata. Counts were checked during extraction and are recorded per collection.

Catalog structure

Links and asset paths are relative to the JSON document that contains them. The catalog currently has no self link. Clients must retain the location from which each document was loaded.

Do not infer undocumented columns, code meanings, or join keys. Collection AGENTS.md files contain only source statements and measurements from the data.

Agent guide for the Moldova geodata catalog

Despre catalog / О каталоге / About the catalog

RO. Catalogul conține date geospațiale selectate din geodata.gov.md. Româna este limba de referință. Textele în rusă și engleză sunt traduceri.

EN. This catalog contains selected geospatial data from geodata.gov.md. Romanian is authoritative; Russian and English are translations.

Data access

DuckDB may interpret EPSG:4326 in latitude-longitude order. Use OGC:CRS84 for longitude-latitude order, or set geometry_always_xy = true.

Spatial filtering

Files are Hilbert-ordered and include a bbox struct with GeoParquet 1.1 covering metadata. DuckDB does not add the bounding-box filter automatically. Apply it before the exact spatial predicate:

The coordinates are metres; this example uses a 10 km square. Measured catalog queries were 3–7 times faster with the explicit bbox filter. Performance varies by collection and query.

Joins

Use cadastru_data_uat2 for aggregation by second-level administrative unit. It contains 35 units and uses the same CRS as the other downloadable collections.

Prefer spatial joins. Fields such as gid and objectid are identifiers within a source layer. The source does not document them as foreign keys between layers.

Base map

maps_base_map_2017 combines source layers that share a schema. Use layer_name for the feature class and source_layer for the exact WFS layer. It has separate point, line, and area assets.

The source has incorrect layer_name values on part of the spot-height layer. Use source_layer when exact source membership matters.

Data limitations

The catalog is curated. Absence from this catalog does not prove absence from the source portal.
Coded columns are preserved without interpretation when the source provides no code list.
Canonical source titles may contain spelling errors. Corrected translations are stored in the language-specific fields.
Canonical language varies: maps uses English and the other workspaces use Romanian. Select a language-specific field when a fixed language is required.
Some source rows have null geometry. Filter with geometry IS NOT NULL before spatial operations.
The GeoServer backends return inconsistent layer metadata. Counts were checked during extraction and are recorded per collection.

Catalog structure

Links and asset paths are relative to the JSON document that contains them. The catalog currently has no self link. Clients must retain the location from which each document was loaded.

Do not infer undocumented columns, code meanings, or join keys. Collection AGENTS.md files contain only source statements and measurements from the data.

1INSTALL spatial; LOAD spatial;
2INSTALL httpfs; LOAD httpfs;
3
4SELECT *
5FROM read_parquet(
6  'https://data.source.coop/nlebovits/moldova-geodata/'
7  || 'cadastru/uat2/cadastru_data_uat2.parquet'
8)
9LIMIT 10;

1INSTALL spatial; LOAD spatial;
2INSTALL httpfs; LOAD httpfs;
3
4SELECT *
5FROM read_parquet(
6  'https://data.source.coop/nlebovits/moldova-geodata/'
7  || 'cadastru/uat2/cadastru_data_uat2.parquet'
8)
9LIMIT 10;

1SET geometry_always_xy = true;
2
3SELECT ST_Transform(geometry, 'EPSG:4026', 'OGC:CRS84')
4FROM read_parquet('cadastru/uat2/cadastru_data_uat2.parquet');

1SET geometry_always_xy = true;
2
3SELECT ST_Transform(geometry, 'EPSG:4026', 'OGC:CRS84')
4FROM read_parquet('cadastru/uat2/cadastru_data_uat2.parquet');

1SELECT count(*)
2FROM read_parquet('cadastru/cladiri/cadastru_data_cladiri.parquet')
3WHERE bbox.xmin >= 200000 AND bbox.xmax <= 210000
4  AND bbox.ymin >= 180000 AND bbox.ymax <= 190000
5  AND ST_Within(
6        geometry,
7        ST_MakeEnvelope(200000, 180000, 210000, 190000)
8      );

1SELECT count(*)
2FROM read_parquet('cadastru/cladiri/cadastru_data_cladiri.parquet')
3WHERE bbox.xmin >= 200000 AND bbox.xmax <= 210000
4  AND bbox.ymin >= 180000 AND bbox.ymax <= 190000
5  AND ST_Within(
6        geometry,
7        ST_MakeEnvelope(200000, 180000, 210000, 190000)
8      );

1WITH units AS (
2  SELECT name, geometry,
3         ST_XMin(geometry) AS x0, ST_YMin(geometry) AS y0,
4         ST_XMax(geometry) AS x1, ST_YMax(geometry) AS y1
5  FROM read_parquet('cadastru/uat2/cadastru_data_uat2.parquet')
6)
7SELECT u.name, count(*) AS buildings
8FROM read_parquet('cadastru/cladiri/cadastru_data_cladiri.parquet') AS b
9JOIN units AS u
10  ON b.bbox.xmin BETWEEN u.x0 AND u.x1
11 AND b.bbox.ymin BETWEEN u.y0 AND u.y1
12 AND ST_Within(ST_Centroid(b.geometry), u.geometry)
13GROUP BY u.name
14ORDER BY buildings DESC;

1WITH units AS (
2  SELECT name, geometry,
3         ST_XMin(geometry) AS x0, ST_YMin(geometry) AS y0,
4         ST_XMax(geometry) AS x1, ST_YMax(geometry) AS y1
5  FROM read_parquet('cadastru/uat2/cadastru_data_uat2.parquet')
6)
7SELECT u.name, count(*) AS buildings
8FROM read_parquet('cadastru/cladiri/cadastru_data_cladiri.parquet') AS b
9JOIN units AS u
10  ON b.bbox.xmin BETWEEN u.x0 AND u.x1
11 AND b.bbox.ymin BETWEEN u.y0 AND u.y1
12 AND ST_Within(ST_Centroid(b.geometry), u.geometry)
13GROUP BY u.name
14ORDER BY buildings DESC;

1SELECT layer_name, count(*) AS features
2FROM read_parquet('maps/base_map_2017/line/*/*.parquet')
3GROUP BY layer_name
4ORDER BY features DESC;

1SELECT layer_name, count(*) AS features
2FROM read_parquet('maps/base_map_2017/line/*/*.parquet')
3GROUP BY layer_name
4ORDER BY features DESC;