GeoPandas.Net.Cesium
0.5.0
dotnet add package GeoPandas.Net.Cesium --version 0.5.0
NuGet\Install-Package GeoPandas.Net.Cesium -Version 0.5.0
<PackageReference Include="GeoPandas.Net.Cesium" Version="0.5.0" />
<PackageVersion Include="GeoPandas.Net.Cesium" Version="0.5.0" />
<PackageReference Include="GeoPandas.Net.Cesium" />
paket add GeoPandas.Net.Cesium --version 0.5.0
#r "nuget: GeoPandas.Net.Cesium, 0.5.0"
#:package GeoPandas.Net.Cesium@0.5.0
#addin nuget:?package=GeoPandas.Net.Cesium&version=0.5.0
#tool nuget:?package=GeoPandas.Net.Cesium&version=0.5.0
<p align="center"><img src="assets/icon.png" width="112" alt="GeoPandas.Net logo"></p>
<h1 align="center">GeoPandas.Net</h1> <p align="center"><b>Next-generation geospatial analytics for .NET 10</b><br> GeoDataFrame · CRS engine · spatial index · GeoParquet / FlatGeobuf / GeoPackage / Shapefile / NetCDF / LAS · PostGIS · lazy queries · spatial statistics · routing · rasters · GeoLLM</p> <p align="center"><a href="README.id.md">🇮🇩 Baca dalam Bahasa Indonesia</a> · <a href="docs/en/index.md">Documentation</a> · <a href="PLAN.md">Roadmap</a> · <a href="Progress.md">Progress</a></p>
Dibuat oleh Gravicode Studios dipimpin oleh Kang Fadhil — Made by Gravicode Studios, led by Kang Fadhil.
GeoPandas.Net re-implements and extends the ideas of Python's GeoPandas for the .NET ecosystem. Like GeoPandas sits on Shapely/GEOS, GeoPandas.Net uses NetTopologySuite (the .NET port of JTS/GEOS) for robust geometry predicates and overlays. Everything above that is implemented here: a column-store GeoDataFrame, a CRS/projection engine with no PROJ dependency, STR-tree/KD-tree/quadtree/octree indexes, file formats, spatial statistics, network analysis, raster/terrain analytics, CAD conversion, an LLM assistant, and Rust + SIMD kernels.
Highlights
| Area | What you get |
|---|---|
| GeoDataFrame | Immutable column store with typed columns and null bitmaps; Filter, SortBy, GroupBy().Agg, Merge, Concat, Describe, ValueCounts; geometry-aware Head/Select/WithColumn |
| Geometry ops | Area, GeomLength, Buffer, Simplify, Centroid, ConvexHull, MakeValid, affine ops, vectorised predicates, all parallel for large inputs |
| Spatial joins | SJoin (intersects / within / contains / touches / crosses / overlaps / covers…), SJoinNearest (exact branch-and-bound), Overlay (intersection, union, difference, symmetric difference, identity), Dissolve, Clip, Explode, Cx |
| CRS engine | The full EPSG dataset (5 800+ CRSs with areas of use), 20 projection methods (TM, LCC, Albers, LAEA, polar & oblique stereographic, Hotine/RSO, Cassini, Equal Earth, Robinson, Mollweide…) matching PROJ to the nanometre, PROJ.4 / WKT1 / ESRI / WKT2 / PROJJSON, Helmert and NTv2 datum shifts, geoids (GTX, ISG, GeoTIFF), EstimateUtmCrs |
| Surveying & GNSS | Traverses (Bowditch/Transit), COGO, least-squares network and leveling adjustment with error ellipses, χ² test and data snooping, RINEX 2/3 and GPS single point positioning |
| Spatial index | STR-tree (bulk-loaded, flat arrays, k-NN), KD-tree, dynamic quadtree, octree (3-D / point clouds), GeoHash; H3 and S2 cell indexes (exact against h3-py and s2sphere) |
| IO | GeoJSON, ESRI Shapefile (.shp/.shx/.dbf/.prj/.cpg, Z), GeoPackage, FlatGeobuf (Hilbert R-tree), GeoParquet 1.1 and Arrow IPC/Feather (GeoArrow, PROJJSON), CSV, DXF behind one GeoIO driver registry; all checked against GDAL / geopandas / pyarrow in both directions |
| Scientific data | NetCDF CDF-1/2/5 + NetCDF-4/HDF5 with CF conventions, hyperslab windows and curvilinear grids; LAS 1.0–1.4 / LAZ (formats 0–10) / COPC point clouds with Octree, DSM/DTM/nDSM; Landsat C2, Sentinel-2 SAFE, STAC, MODIS/VIIRS product readers |
| Databases | ReadSql / ReadTable / ToSql (GeoPandas read_postgis/to_postgis) for PostGIS, SQL Server, DuckDB Spatial, Oracle Spatial, MySQL over plain ADO.NET, plus BigQuery GIS |
| Lazy & versioned | LazyGeoDataFrame: bbox/column/attribute/limit pushdown into files and SQL, streaming batches, Explain(); DatasetRepository: content-addressed snapshots, commits, tags, keyed diffs |
| Analytics | Queen/rook/KNN/distance-band weights, Moran's I, Geary's C, LISA, Getis-Ord Gi* hot spots, DBSCAN, K-Means, IDW, ordinary and universal kriging, OLS with LM diagnostics, ML spatial lag/error models, GWR (matching PySAL spreg/mgwr) |
| Network | Graphs from lines (costs, one-way), Dijkstra, A*, bidirectional Dijkstra, Contraction Hierarchies, service areas / isochrones, connected components, utility-network tracing; water hydraulics (EPANET .inp, global gradient solver matching EPANET 2.2), topology rules, isolation-valve segments |
| Civil & BIM | Horizontal (clothoid) and vertical alignments, corridors daylighted to a DEM, cut/fill and mass haul, level pads, parcel subdivision; IFC2x3/IFC4 → GeoDataFrame in map coordinates, LOD1 footprints, GIS → georeferenced IFC4 |
| Raster | GeoTIFF / BigTIFF / COG reader (strips, tiles, Deflate, LZW, PackBits, predictor), GeoTIFF writer, slope, aspect, hillshade, ruggedness, viewshed, sink filling, D8 flow, accumulation, watersheds, NDVI/NDWI/NDBI/SAVI, Otsu change detection, zonal stats, rasterize, contours |
| CAD | DXF reader (LINE, LWPOLYLINE with bulges, POLYLINE, CIRCLE, ARC, ELLIPSE, SPLINE, TEXT, MTEXT, INSERT, 3DFACE) and R12 writer, DWG read/write (R14–2018), MicroStation DGN v7 read/write, two-point CAD → GIS georeferencing |
| Spatial AI | GeoAssistant (GeoLLM): natural language → whitelisted query plan, executed deterministically. Works with OpenAI, Azure OpenAI, DeepSeek or any OpenAI-compatible endpoint |
| Scale | PartitionedGeoDataFrame (k-d tree, grid, H3 or attribute partitions; parallel map/reduce; partition-aware spatial join identical to SJoin) and partitioned GeoParquet datasets with bounding-box partition pruning |
| 3D & twins | glTF/GLB and OGC 3D Tiles 1.1 (extruded buildings, terrain; validator-clean, ECEF-georeferenced), Cesium.Net integration, interactive Explore() maps (MapLibre, Blazor component), live IoT layers with geofence and threshold events |
| Cloud, ML & GPU | Azure Maps, AWS Location (SigV4) and Azure Digital Twins clients returning GeoDataFrames; ONNX segmentation and ML.NET land-cover classification; ILGPU raster kernels (CUDA / OpenCL / CPU) |
| Maps | gdf.Plot() returns a renderer-agnostic MapScene (choropleth, categorical, proportional symbols, labels, legends, scale bar, north arrow, map-sheet frame) that can be drawn as SVG, in Blazor, or natively in Avalonia |
| Performance | Rust native kernels (Rayon) with a Vector<T> SIMD managed fallback, CSR graphs, Span<T>/ArrayPool<T>, AOT-compatible core packages |
Quick start
dotnet add package GeoPandas.Net
using GeoPandas.Net;
using GeoPandas.Net.Data;
using GeoPandas.Net.Geometry;
using GeoPandas.Net.Plotting;
var districts = Gpd.ReadFile("districts.gpkg"); // GeoJSON, .shp, .gpkg, .csv, .dxf
var schools = Gpd.ReadFile("schools.geojson").ToCrs(districts.Crs!);
// Spatial join + group by
var perDistrict = schools.SJoin(districts, SpatialPredicate.Within)
.GroupBy("name_right")
.Agg(("name_left", AggFunc.Count, "schools"));
// Reproject to metres, measure, buffer
var utm = districts.ToCrs(districts.EstimateUtmCrs());
var area = utm.Area(); // m²
var walk = schools.ToCrs(utm.Crs!).Buffer(500).Dissolve(); // 500 m walking catchment
// Map → SVG
utm.WithColumn(area.Rename("area_m2"))
.Plot(new LayerStyle { Column = "area_m2", ColorMap = ColorMaps.Viridis }, "District area")
.SaveSvg("area.svg");
utm.ToFile("districts_utm.shp");
More examples: Getting started · GeoDataFrame & IO · Formats, databases, lazy & versioning · Analysis · CAD & AI · Engineering & BIM · Scale, 3D, twins & cloud.
Packages
| Package | Contents |
|---|---|
GeoPandas.Net |
Meta-package with everything below, plus the Gpd facade |
GeoPandas.Net.Geometry |
Geometry factory, CRS engine, spatial indexes, geodesy, native kernels |
GeoPandas.Net.Core |
DataFrame, GeoSeries, GeoDataFrame, spatial joins/overlay, MapScene plotting |
GeoPandas.Net.Data |
GeoJSON, Shapefile, GeoPackage, FlatGeobuf, CSV, GeoIO registry and GeoIO.Scan, dataset versioning |
GeoPandas.Net.Arrow |
GeoParquet 1.1 reader/writer, Arrow IPC / Feather, DataFrame ⇄ RecordBatch |
GeoPandas.Net.PointCloud |
LAS/LAZ/COPC, PointCloud (Octree, classes, DSM/DTM, GeoDataFrame interop) |
GeoPandas.Net.Survey |
Traverses, COGO, least-squares adjustment, RINEX, GNSS single point positioning |
GeoPandas.Net.RemoteSensing |
Landsat C2, Sentinel-2 SAFE (JPEG 2000), STAC items, MODIS/VIIRS grids and HDF-EOS5 |
GeoPandas.Net.Sql |
PostGIS / SQL Server / DuckDB / Oracle / MySQL dialects, ReadSql, ToSql, SQL scan source |
GeoPandas.Net.Sql.BigQuery |
BigQuery GIS read / load (separate, brings Google.Cloud.BigQuery.V2) |
GeoPandas.Net.Analytics |
Spatial weights, autocorrelation, hot spots, clustering, interpolation, spatial regression, GWR |
GeoPandas.Net.Network |
Routing, Contraction Hierarchies, isochrones, utility tracing, water hydraulics, topology rules, valve isolation |
GeoPandas.Net.Civil |
Alignments, corridors, earthwork, mass haul, parcel subdivision |
GeoPandas.Net.BIM |
STEP/IFC reader and writer, IFC ⇄ GIS, LOD1 building footprints |
GeoPandas.Net.Raster |
GeoTIFF/COG IO, NetCDF (classic + NetCDF-4), terrain, hydrology, spectral indices, zonal statistics |
GeoPandas.Net.CAD |
DXF, DWG and DGN v7 read/write, CAD ⇄ GIS georeferencing |
GeoPandas.Net.AI |
GeoLLM assistant, OpenAI / Azure OpenAI client |
GeoPandas.Net.Scene |
Local ENU frames, meshes (terrain, extrusions, points, lines), glTF/GLB with feature metadata, 3D Tiles 1.1 |
GeoPandas.Net.Twin |
Live layers for IoT streams, geofence/threshold/stale events, zone statistics, digital twins |
GeoPandas.Net.Cloud |
Azure Maps, AWS Location Service, Azure Digital Twins (no SDK dependencies) |
GeoPandas.Net.Cesium |
Cesium.Net helpers: GeoDataFrames, point clouds and tilesets in a CesiumJS viewer (separate package) |
GeoPandas.Net.Blazor |
ExploreView (interactive MapLibre map) and MapSceneView components (separate package) |
GeoPandas.Net.ML |
ONNX Runtime segmentation, ML.NET land-cover classification, accuracy assessment (separate package) |
GeoPandas.Net.Gpu |
ILGPU raster kernels: slope, aspect, hillshade, normalized difference, convolution, IDW (separate package) |
GeoPandas.Net.Datasets |
Reproducible sample data (city, roads, POIs, DEM, imagery, CAD plan, LiDAR tile, climate grid, world cities) |
GeoPandas.Net.Cli |
gpd command-line tool (dotnet tool install -g GeoPandas.Net.Cli) |
GeoPandas.Net.Templates |
dotnet new gpd-console, gpd-blazor, gpd-avalonia |
Gallery (Blazor web + Avalonia desktop)
36 live samples across GIS, data & formats, spatial statistics, networks, raster, civil engineering, remote sensing, surveying & geodesy, CAD & BIM, digital twins & 3D and AI. The web and desktop galleries render from the same scene display list, so their maps are identical. The UI is bilingual (EN/ID).
dotnet run --project gallery/GeoPandas.Net.Gallery.Blazor # http://localhost:5280
dotnet run --project gallery/GeoPandas.Net.Gallery.Avalonia
gdf.Explore() — interactive MapLibre map with hover tooltips and legend:
Avalonia desktop, including the Phase 2 data & format samples:
CLI
dotnet tool install -g GeoPandas.Net.Cli
gpd sample districts districts.gpkg
gpd info districts.gpkg
gpd convert districts.gpkg districts.geojson --to-crs EPSG:4326
gpd sjoin pois.geojson districts.gpkg joined.gpkg --predicate within
gpd plot districts.gpkg map.svg --column population --cmap YlOrRd
gpd route roads.shp --from 700000,9310000 --to 705000,9315000 --out route.geojson
gpd raster hillshade dem.tif hillshade.tif
gpd convert districts.gpkg districts.parquet # GeoParquet / .fgb / .feather
gpd scan roads.parquet --bbox 700000,9308000,704000,9312000 --where "highway=primary" --explain
gpd las ndsm tile.laz heights.tif --cell 1
gpd netcdf raster era5.nc t2m t2m.tif --index 0
gpd repo commit parcels.repo parcels.gpkg --m "weekly sync"
gpd ask districts.gpkg "5 distrik terpadat di wilayah Selatan?"
See docs/en/tools.md for every command.
Building from source
Requirements: .NET SDK 10, plus Rust (cargo) if you want to build the optional native kernels.
cargo build --release --manifest-path native/geopandas_native/Cargo.toml # optional, enables the "rust" backend
dotnet build GeoPandas.Net.slnx
dotnet test tests/GeoPandas.Net.Tests
dotnet run -c Release --project benchmarks/GeoPandas.Net.Benchmarks -- --filter "*"
Set GEOPANDAS_NET_DISABLE_NATIVE=1 to force the managed SIMD path.
License & credits
MIT. Geometry engine: NetTopologySuite (BSD-3). Dibuat oleh Gravicode Studios dipimpin oleh Kang Fadhil.
| Product | Versions Compatible and additional computed target framework versions. |
|---|---|
| .NET | net10.0 is compatible. net10.0-android was computed. net10.0-browser was computed. net10.0-ios was computed. net10.0-maccatalyst was computed. net10.0-macos was computed. net10.0-tvos was computed. net10.0-windows was computed. |
-
net10.0
- Cesium.Net (>= 1.0.0)
- GeoPandas.Net.Data (>= 0.5.0)
- GeoPandas.Net.Scene (>= 0.5.0)
NuGet packages
This package is not used by any NuGet packages.
GitHub repositories
This package is not used by any popular GitHub repositories.
| Version | Downloads | Last Updated |
|---|---|---|
| 0.5.0 | 0 | 11/18/2026 |