nettopologysuite

name	nettopologysuite
description	Use when performing 2D computational geometry in .NET — spatial predicates, overlay operations, buffering, WKT/WKB I/O. NetTopologySuite: the .NET port of JTS, available via NuGet for spatial analysis in C# applications.
tags	["dotnet","csharp","geometry","spatial","wkt","wkb","geojson","ef-core","npgsql"]

项目地址： https://github.com/NetTopologySuite/NetTopologySuite

NuGet： NetTopologySuite

官方文档： https://nettopologysuite.github.io/NetTopologySuite/

许可证： BSD-3-Clause

概述

NTS 与 JTS API 几乎一一对应（C# 命名风格）：

• 几何模型：Point/LineString/Polygon/Multi*/GeometryCollection
• 空间关系：Equals/Contains/Within/Intersects/Touches/Crosses/Overlaps/Disjoint/Relate
• 集合运算：Intersection/Union/Difference/SymmetricDifference/Buffer
• 几何分析：ConvexHull/Centroid/Area/Length/Distance/IsValid/Simplify
• 空间索引：STRtree/Quadtree
• 格式读写：WKT、WKB、GeoJSON、Shapefile（独立 NuGet）
• EF Core 集成：SqlServer 与 PostgreSQL 自动翻译为空间 SQL

---

环境准备

dotnet add package NetTopologySuite
dotnet add package NetTopologySuite.IO.GeoJSON
dotnet add package NetTopologySuite.IO.ShapeFile
dotnet add package ProjNet                          # 坐标变换（可选）

.NET 6+ 推荐。

---

几何工厂

using NetTopologySuite.Geometries;

var gf = new GeometryFactory(new PrecisionModel(), 4326);

var p  = gf.CreatePoint(new Coordinate(116.397, 39.908));
var ls = gf.CreateLineString(new[] {
    new Coordinate(116, 39), new Coordinate(117, 40)
});
var poly = gf.CreatePolygon(new[] {
    new Coordinate(0,0), new Coordinate(10,0),
    new Coordinate(10,10), new Coordinate(0,10), new Coordinate(0,0)
});

---

空间关系与运算

bool b   = polyA.Intersects(polyB);
var inter = polyA.Intersection(polyB);
var union = polyA.Union(polyB);
var diff  = polyA.Difference(polyB);

var buf  = p.Buffer(100);
var hull = mp.ConvexHull();
double area = poly.Area;
double len  = ls.Length;
double dist = polyA.Distance(polyB);

string rel = polyA.Relate(polyB).ToString();
bool ok    = polyA.Relate(polyB, "T*F**F***");

---

WKT / WKB / GeoJSON

using NetTopologySuite.IO;

var rdr = new WKTReader();
var g = rdr.Read("POINT (116.4 39.9)");

string wkt = new WKTWriter().Write(g);
byte[] wkb = new WKBWriter().Write(g);

using NetTopologySuite.IO.Converters;
using System.Text.Json;
var opts = new JsonSerializerOptions();
opts.Converters.Add(new GeoJsonConverterFactory());
string json = JsonSerializer.Serialize(g, opts);
var g2 = JsonSerializer.Deserialize<Geometry>(json, opts);

---

Shapefile

using NetTopologySuite.IO.Esri;
foreach (var f in Shapefile.ReadAllFeatures("input.shp"))
    Console.WriteLine($"{f.Geometry} {f.Attributes["NAME"]}");

Shapefile.WriteAllFeatures(features, "output.shp");

---

空间索引

using NetTopologySuite.Index.Strtree;

var tree = new STRtree<Geometry>();
foreach (var g in geometries)
    tree.Insert(g.EnvelopeInternal, g);
tree.Build();

var hits   = tree.Query(queryEnv);
var nearby = tree.NearestNeighbour(p.EnvelopeInternal, p, new GeometryItemDistance());

---

几何修复与简化

using NetTopologySuite.Operation.Valid;
using NetTopologySuite.Simplify;
using NetTopologySuite.Geometries.Utilities;

if (!poly.IsValid)
    poly = (Polygon)GeometryFixer.Fix(poly);

var simp = TopologyPreservingSimplifier.Simplify(poly, 0.001);

---

EF Core + SQL Server / PostgreSQL

dotnet add package Microsoft.EntityFrameworkCore.SqlServer.NetTopologySuite
# 或
dotnet add package Npgsql.EntityFrameworkCore.PostgreSQL.NetTopologySuite

public class City {
    public int Id { get; set; }
    public string Name { get; set; }
    public Point Location { get; set; }
    public Geometry Boundary { get; set; }
}

services.AddDbContext<AppDb>(o =>
    o.UseSqlServer(conn, x => x.UseNetTopologySuite()));

var nearby = db.Cities
    .Where(c => c.Location.Distance(target) < 5000)
    .OrderBy(c => c.Location.Distance(target))
    .Take(10)
    .ToList();   // 自动翻译为空间 SQL

---

坐标系转换（ProjNet）

using ProjNet.CoordinateSystems;
using ProjNet.CoordinateSystems.Transformations;

var src = GeographicCoordinateSystem.WGS84;
var tgt = ProjectedCoordinateSystem.WebMercator;
var ct  = new CoordinateTransformationFactory()
            .CreateFromCoordinateSystems(src, tgt);
double[] xy = ct.MathTransform.Transform(new[] { 116.397, 39.908 });

---

性能优化

1. 共享 GeometryFactory，固定 SRID
2. 大批量空间查询用 STRtree
3. WKB 比 WKT 更快更紧凑
4. 多次空间关系判断使用 PreparedGeometry：

using NetTopologySuite.Geometries.Prepared;
var prep = PreparedGeometryFactory.Prepare(largePoly);
foreach (var p in points) if (prep.Contains(p)) ...

---

常见问题

问题	解决
TopologyException	GeometryFixer.Fix 修复无效几何
EF Core 类型冲突	用 UseNetTopologySuite() 而非 Microsoft.SqlServer.Types
Z/M 坐标丢失	配置 WKBWriter 的 HandleOrdinates
Buffer 慢	减小 quadrantSegments、并行处理

---

AI 使用建议

推荐工作流

1. 创建几何工厂：使用 new GeometryFactory(new PrecisionModel(), 4326) 创建工厂，SRID 指定坐标系
2. 创建几何对象：通过工厂方法 CreatePoint()、CreateLineString()、CreatePolygon() 创建几何
3. 空间运算：直接调用几何对象方法 Intersects()、Intersection()、Buffer() 等
4. 格式读写：使用 WKTReader/WKTWriter、GeoJsonConverterFactory 进行序列化
5. EF Core 集成：UseNetTopologySuite() 启用空间查询翻译为 SQL

关键注意事项

• GeometryFactory 共享：固定 SRID，复用 GeometryFactory 实例
• 拓扑异常修复：无效几何使用 GeometryFixer.Fix() 修复
• PreparedGeometry：重复空间关系判断时使用 PreparedGeometryFactory.Prepare() 大幅提速
• STRtree 批量查询：先 Build() 再 Query()，大数据量场景优于逐几何判断
• WKB vs WKT：WKB 比 WKT 更快更紧凑，数据库交互优先使用 WKB
• Z/M 坐标：WKBWriter 需配置 HandleOrdinates 保留 Z/M 坐标

相关技能

• jts — JTS Topology Suite（Java 原版）：../jts/SKILL.md
• geometry-api-net — Esri Geometry API for .NET：../geometry-api-net/SKILL.md
• mapsui — .NET 跨平台地图组件：../mapsui/SKILL.md
• opengis-utils-for-net — .NET GIS 统一工具包：../opengis-utils-for-net/SKILL.md

参考资源

• 文档：https://nettopologysuite.github.io/NetTopologySuite/
• API：https://nettopologysuite.github.io/NetTopologySuite/api/
• JTS Javadoc（API 等价）：https://locationtech.github.io/jts/javadoc
• 中文教程（znlgis）：https://znlgis.github.io/gis/tutorial/nettopologysuite/