Meshes Providers – Guide & Recipes¶

Applies to MIKECloud (Multidimensional datasets) and MIKECore/DFSU (local files). Use this to pick a provider fast, wire it, and avoid foot-guns.

Quick map of the providers (big picture)¶

Provider	Type	“Groups” meaning	Where data lives	Highlights
`DHI.Services.Provider.MIKECloud.GroupedMeshRepository`	Grouped	MIKECloud Projects/Subprojects	MIKECloud Multidimensional sets	Discover metadata, timestamps, point values, aggregations (whole/polygon/many), import dfs2/dfsu
`DHI.Services.Provider.MIKECore.DfsuMeshRepository`	Grouped	File system folders	Local/remote DFSU files	Discover, timestamps, point values, aggregations, contouring (mesh + element values)

Both expose the same Meshes service surface (discover, timestamps, point series, aggregations).

Which one should I pick?¶

Your results are already in MIKECloud or you want cloud-side compute + import → MIKECloud
You have raw DFSU files on disk/object storage and need contours too → DFSU

Capabilities at a glance¶

Capability / API	MIKECloud	DFSU
Discover meshes / read `MeshInfo`	✓	✓
`GetDateTimes(fullName)`	✓	✓
Point values → `ITimeSeriesData<double>`	✓	✓
Aggregations (mesh / polygon / many polygons)	✓	✓
Instantaneous aggregation (single timestamp)	✓	✓
Import dfs2/dfsu streams	✓	—
Contours / raw snapshot (`GetMeshData`)	✗	✓

CRS rules

MIKECloud: pass WGS84 (EPSG:4326) for Point/Polygon.
DFSU: pass lon/lat; provider projects to mesh CRS internally. Z on points not supported.

Time rules

MIKECloud: timestamps come from a regular temporal domain → always use exact values returned by GetDateTimes.
DFSU: same: exact timestamps from GetDateTimes.

Registering providers (minimal wiring)¶

A) MIKECloud (grouped)¶

using DHI.Services.Meshes;
using DHI.Services.Provider.MIKECloud;

var repo = new GroupedMeshRepository("apiKey=<KEY>;projectId=<PROJECT>;environment=Prod");
ServiceLocator.Register(new GroupedMeshService<string>(repo), "mc-meshes"); // connectionId

B) DFSU (grouped)¶

using DHI.Services.Meshes;
using DHI.Services.Provider.MIKECore;

var fs   = new FileSource(@"C:\data\meshes");            // or your IFileSource
var repo = new DfsuMeshRepository(fs, filePathOrPrefix: "dfsu");
ServiceLocator.Register(new GroupedMeshService<string>(repo), "dfsu");      // connectionId

After registration your host exposes /api/meshes/{connectionId}/… routes (per your Meshes Web API).

IDs & groups (mental model)¶

MIKECloud: "Group/Subgroup/DatasetName" maps to Subprojects under a root project.
DFSU: Full file path (under your filePathOrPrefix) becomes the Id; group is the relative folder.
Groups are hierarchical: you can list by group (recursive) and use tree-options if your host exposes them.

Copy-paste recipes¶

A) MIKECloud — import, timestamps, point values, aggregations¶

var svc = Services.Get<GroupedMeshService<string>>("mc-meshes");

// Import dfs2/dfsu stream
using var fs = File.OpenRead(@"C:\data\Result.dfs2");
svc.Repository.Add(fs, "Models/Baltic/Result.dfs2");

// Timestamps
var id    = "Models/Baltic/Result.dfs2";
var times = svc.Repository.GetDateTimes(id).ToArray();

// Point series (lon,lat in WGS84)
var p     = new DHI.Services.Spatial.Point(12.34, 55.67);
var range = new DHI.Services.TimeSeries.DateRange(times.First(), times.First().AddDays(2));
var wl    = svc.Repository.GetValues(id, "Water Level", p, range);

// Aggregations
var poly = DHI.Services.Spatial.Polygon.FromWKT("POLYGON((12 55,12.5 55,12.5 55.5,12 55.5,12 55))");
var avg  = svc.Repository.GetAggregatedValues(id, AggregationType.Average, "Water Level", poly, range);

B) DFSU — timestamps, point values, polygon/mesh aggregates, contours¶

var svc   = Services.Get<GroupedMeshService<string>>("dfsu");
var id    = @"Scenarios/Coast/mesh01.dfsu";                 // relative under your prefix
var times = svc.Repository.GetDateTimes(id).ToArray();

// Point values
var point = new Spatial.Point(12.345, 55.678);              // lon,lat
var dr    = new DateRange(times[0], times[0].AddHours(6));
var ts    = svc.Repository.GetValues(id, "Water Level", point, dr);

// Whole-mesh max at an instant
var maxAt = svc.Repository.GetAggregatedValue(id, AggregationType.Maximum, "Water Level", times[0]);

// Many polygons at an instant (order preserved)
var polys = new [] {
  Spatial.Polygon.FromWKT("POLYGON((12 55,12.5 55,12.5 55.5,12 55.5,12 55))"),
  Spatial.Polygon.FromWKT("POLYGON((12.6 55.2,12.8 55.2,12.8 55.4,12.6 55.4,12.6 55.2))")
};
var many  = svc.Repository.GetAggregatedValues(id, AggregationType.Average, "Water Level", polys, times[1]);

// Contours (DFSU only): mesh + element values → build in service layer
var fc = svc.GetContours(id, "Water Level", new double[] { -1, 0, 1, 2 }, times[2]);

What you’ll actually call (service surface)¶

Discovery: GetAll(), GetByGroup(group), GetFullNames(), ContainsGroup(group)
Metadata: Get(fullName) → MeshInfo<string> (items, units, WKT/projection, bbox, date range)
Time: GetDateTimes(fullName) → timestamps (ascending)
Point values: GetValues(fullName, item, Point, DateRange)
Aggregations:
- Whole mesh: GetAggregatedValues(fullName, aggregation, item, DateRange) / GetAggregatedValue(..., dateTime)
- Polygon(s): GetAggregatedValues(fullName, aggregation, item, Polygon|IEnumerable<Polygon>, DateRange|DateTime)
Import/remove (MIKECloud only): Add(Stream, fullName), Remove(fullName)
Contours (DFSU only): GetMeshData (repo) → GetContours (service convenience)

Foot-guns & best practices¶

Time alignment

Always use timestamps returned by GetDateTimes(id). Both providers index by exact match.

CRS

MIKECloud expects WGS84 geometries; DFSU accepts lon/lat and projects internally.
Z on points is not supported in DFSU point extraction.

MIKECloud

No contours: GetMeshData is not implemented.
If a vertical domain exists, queries use layer index 0.

DFSU

Polygon–mesh intersection can be expensive; the repo precomputes weights once per request and reuses them.
Missing element values (delete value) are filtered before aggregations; for point series they’re kept as-is.
For “all items” time series, item order is by item number to minimize file seeking.

Performance

Cache GetDateTimes(id) per dataset.
Prefer batch polygon aggregation for many polygons at one instant (dedicated API).

TL;DR

MIKECloud → cloud datasets, imports, point/area analytics (no contours).
DFSU → local files, same analytics plus contours.
Both are grouped; register in code; drive everything via GroupedMeshService<string>.