Documentation

Platform v3.0.0 Docs v3.0.0

Terranoetis is a real-time geospatial intelligence platform: a CesiumJS globe, an Express/TypeScript API, 150 literature-cited analytical equation engines, seven hazard-physics simulation kernels, and an AI chat pipeline with dual-process cognition. These documents state what the software does — verified against the source in this repository, and against the code actually executed wherever this environment allows.

Accuracy policy

Physics statements carry a file:line citation into the source. Executed evidence is tagged MEASURED. See Methodology & verification and its discrepancy log.

Verified at a glance

Analytical engines: 150 (ids 1–150, no gaps) across 7 parts and 26 domains — counted programmatically src/data/analyticalModels.ts:801+server/analytical-models/engine.ts:1-174
Equation-engine tests: 1,653 unit tests passed (141 files, 22.85 s wall) with npm run test:unit measured
Integration tests: 49 passed (5 files, 5.66 s) with npm run test:integration measured
Simulation kernels: 7 Python kernels; 3 execute locally on CPU (earthquake, wildfire, hurricane); 4 dispatch through Kaggle. All are NumPy-only (0 cupy/CUDA imports); only flood-sim requests the GPU accelerator in its metadata. server/kaggle/simRunner.ts:126-141 kaggle-kernels/*/kernel-metadata.json
Hazard physics: All 7 kernels executed locally in this review; representative + boundary inputs, conservation gates, convergence and benchmark suites re-run — see Evidence runs measured
Citations: 150 model reference strings; 80 contain a DOI, 12 an ISBN, 4 flagged NO-DOI; spot-checked DOI 10.1785/0120130065 resolves (HTTP 302) src/data/analyticalModels.ts (counted)

Where to go next

Getting started

Install, run, verify: the five-minute local setup with measured outputs.

Platform architecture

Three-tier topology, request lifecycle, background services.

Hazard simulations

Seven physics kernels: modes, contracts, measured timings.

Earthquake

BSSA14 GMPE ShakeMap · local CPU · measured 0.008 s compute.

Tsunami

HLL/minmod SWE over GEBCO 2020 · known sloped-bed limitation documented.

Volcano

Lava SWE + Morton–Taylor plume + ash PDE · benchmarks 4/4 PASS.

Landslide

Voellmy–Salm debris flow · convergence gates PASS.

Flood

Local-inertial SWE · mass closure 0.0000 % measured.

Hurricane

Holland wind + SWE surge + waves + rainfall + runoff · gates PASS.

Wildfire

Rothermel/Anderson-13 · K_ROS calibration PASS measured.

Analytical equation engine

150 engines · 7-stage workflow · honest-NaN contract.

AI chat & cognition

Intent routing, System 1 / System 2, 9-provider Omninet router.

Live data & monitoring

82 registered external APIs, Sentinel polling, reflex actions.

World model & causal reasoning

Causal graph, ensemble predictor, forecast ledger, KG v2.

Memory system

Tiered working/episodic/semantic memory with SQLite + Redis hot path.

3D globe visualization

React 19 + Cesium 1.140; 41 rendering modules (counted).

Platform services

Auth, rate limits, sandbox, queue, plugins, observability.

REST & WebSocket API

Endpoint map, auth model, error envelope; canonical list in API.md.

Simulation job API

Parameter contracts, job lifecycle, results formats, UQ endpoints.

Methodology & verification

How every claim on this site was checked; discrepancy log.

Deployment & operations

Docker, environment, CI, optional Kaggle credentials.

Style guide

The writing, citation and accessibility standards these pages follow.

Documentation coverage

Every area of the platform has a long-form narrative page with diagrams, rendered from the Markdown sources of truth in the repository.
Area	Covers	Read
Architecture	System topology, request flow, background services, module inventory	Architecture narrative
API Reference	360+ REST endpoints + WebSocket channel	API narrative
Frontend	React components, rendering engine, simulation overlays, hooks	Frontend narrative
Backend	Server modules, cognition, memory, security, observability	Backend narrative
Analytical Models	The 150 equation engines, 7 parts, 26 domains	Analytical-models narrative
Deployment	Docker, environment variables, Kaggle token setup, CI/CD, security	Deployment narrative

Repository map

Top-level layout
Path	Contents
`src/`	React 19 + CesiumJS client (App shell, components/, rendering/ 41 modules, hooks, lib, services, store)
`server/`	Express API: index.ts (14,573 lines), agent/cognition, analytical-models, kaggle, sentinel, memory, world-model, middleware, db
`kaggle-kernels/`	Seven Python simulation kernels + per-kernel metadata; runtime output goes to `results/<jobId>/` (gitignored)
`docs/`	This site: generated HTML pages (source: scripts/docs/), Markdown prose of record, manifest and sitemap
`e2e/`	Playwright specs (7 files; 15 test declarations, 14 registered by default — 1 gated behind `LULC_HEAVY=1`)
`scripts/`	Dev utilities, live smoke scripts, docs build + quality gate
`.github/workflows/`	ci.yml (10 jobs) + docs.yml (quality gate)