Vision

TopoBathySim is a policy-driven topobathymetric fusion runtime: a Python library (and optional service) for turning heterogeneous elevation sources into seamless, simulation- and analysis-ready grids with per-pixel provenance.

Mission

Make it straightforward to build trustworthy digital terrain anywhere on Earth by combining global basemaps with local high-resolution surveys—while keeping assumptions explicit and outputs reproducible.

What TopoBathySim is

TopoBathySim focuses on the layer between raw geospatial infrastructure and downstream modeling/simulation.

Concretely, it provides:

  • Providers that fetch and normalize data layers (global grids, regional compilations, local surveys/lidar), with caching for offline replay.

  • Policies (YAML) that encode domain expertise as deterministic fusion strategies per variable (elevation now; time-varying covariates next).

  • Operators for transition zones (e.g., seamline feathering), so outputs can be continuous without hiding where each dataset came from.

  • Provenance-first outputs (xarray.Dataset) that include fused values and source masks for debugging, QA, and downstream trust.

  • An optional microservice that serves the same fused products over HTTP for real-time clients (Unity/Unreal/Omniverse, robotics stacks, web viewers).

What TopoBathySim is not

TopoBathySim is intentionally not:

  • A hosted dataset, or an “authoritative” replacement for hydrographic offices.

  • A single “one true fusion algorithm.” Fusion strategies are use-case dependent and belong in policies.

  • A purely visualization-oriented tool. PNG/hillshade and viewers are debugging tools; data products are the core output.

  • A guarantee of global vertical datum harmonization. Datum assumptions and transforms are surfaced explicitly in metadata/provenance, and regional transformations are pluggable.

Where we fit in the open topo/bathy ecosystem

TopoBathySim complements existing tools and workflows:

  • CUDEM and similar pipelines are excellent for systematic DEM production and gridding workflows.

  • GDAL / PDAL / xarray / rioxarray provide foundational geospatial I/O and processing primitives.

  • bmi-topography provides standardized access patterns to certain DEM products.

TopoBathySim’s niche is the runtime + policy layer:

  • codifying “what should win where” (trusted overwrite vs transition blending),

  • making results reproducible and inspectable (provenance masks),

  • and delivering outputs in forms that simulation and modeling stacks can consume quickly (arrays, tiles, offline caches).

Principles

  1. Overwrite in trusted zones; blend in transition zones Survey “truth” should win where present; seams should be softened where coverage edges meet.

  2. Provenance is first-class Every fused product should carry a source mask and enough metadata to explain how it was produced.

  3. Global-first, regional-extensible A global basemap policy should work anywhere. Regional expertise should be added via provider plugins and policy presets.

  4. Deterministic policies Policies are versionable artifacts (YAML) that can be reviewed, shared, and cited. Natural-language interfaces may exist later, but must compile to deterministic policies.

  5. Data-first outputs The canonical output is an xarray.Dataset. Visualizations exist to validate the data and debug fusion logic.

How to contribute (most valuable paths)

  • Add a provider for a dataset relevant to your region (EU/Asia-Pacific/etc.).

  • Add a policy preset that encodes best practices and local domain knowledge.

  • Add operators or tests that improve seam handling, performance, or QA.

  • Help expand the platform toward time-varying covariates (Sentinel-1/2/3 and beyond), with consistent caching and provenance.

See CONTRIBUTING.md and the policies/ gallery for entry points.