Uploading data
Drop a file, get a layer. Uploads run in the background through a real ingest pipeline, reprojection, indexing, tiling and automatic overviews for large datasets.
Where to upload#
- The Upload button in the content browser (files land in your current folder).
- Upload data on the portal home page.
- The Upload button in the Data Editor’s library rail.
- The map viewer’s Add → Upload tab, the new layer drops straight into the open map.
All of them open the same dialog: drop files or browse, several at once, each file becomes its own background job, and you can keep working while they process. Uploading requires the Editor role or higher.
Supported formats#
| Kind | Formats | Becomes |
|---|---|---|
| Spatial vector | GeoPackage, Shapefile (zipped or .shp set), GeoJSON, KML/KMZ, File Geodatabase, GML, GPX, DXF, MapInfo TAB, FlatGeobuf, GeoParquet | A feature layer (Dataset item) |
| CSV | .csv, with a column-mapping step (see below) | A point layer or an attribute table |
| Spreadsheet | XLSX, ODS (re-save .xls as .xlsx) | An attribute table |
| Files | Images (PNG, JPEG, WebP, SVG, TIFF…), PDF, other documents | Image / PDF / Document items |
Size limits: 5 GB per spatial file or spreadsheet, 200 MB for CSV, 100 MB for images and documents. Executables and active web content are rejected outright.
A GeoPackage or File Geodatabase containing several layers imports its first layer only, there is no layer picker in the web uploader yet. Export single-layer files (or upload via QGIS, one layer at a time).
Coordinate systems are read from the file#
You never pick a CRS at upload, OrbGIS reads it from the file and reprojects as needed. CSV coordinates are treated as WGS 84 latitude/longitude, with a sanity check that fails clearly if the numbers look like projected metres. More in Coordinate reference systems.
CSV uploads#
A single CSV gets a configuration step: choose Point layer and pick the latitude/longitude columns (auto-detected from common names like lat, lon, x, y, with a live “N points found” count), or choose Table (no geometry) to import attributes only, which you can later join to layers via relationships. Upload CSVs one at a time so you get this step.
What happens during processing#
Progress lives in the header’s upload panel. After Uploading… and a spell in the queue you’ll see the pipeline phases: Receiving file… → Reading layer… → Importing features… (with a live feature count) → Optimizing for fast access… → Finalizing layer… → Calculating extent…, and for large layers, an overview-building step. When the job says Done, the layer renders fast at every zoom level.
Large layers get automatic overviews#
Layers past roughly 100,000 features (or with very heavy geometry) get a level-of-detail pyramid built at ingest: zoomed far out you see a density rendering that traces the dataset’s true shape; mid zooms show every feature lightly; close up you get full geometry and attributes with your styling. Selection happens server-side per tile, nothing to configure, and it’s why national-scale datasets stay smooth. If an overview build ever fails, the layer still works from source and shows an “Overview unavailable” chip. Polygon class-grids (like land-cover cells) can additionally opt into mode aggregation from the dataset page, majority class per coarser cell.
Failed jobs can be retried from the upload panel; cancelling after import has begun discards the partial layer. Overwriting an existing layer’s data with a new file goes through the Data Editor.