mud2dust developer docs
How to read mud2dust's public soil-moisture data, and how to contribute sensors, drone flights, lab samples, or field boundaries through the open contribution API.
Tile server
Read soil-moisture + companion rasters
titiler XYZ tile API at mud2dust.io. Use it from Mapbox GL, MapLibre, QGIS, leafmap, or anything that speaks /{z}/{x}/{y}. No signup.
Sensor / partner API
Connect probes, weather stations, drone flights, lab samples
Versioned at /v1, live at api.mud2dust.io. OpenAPI 3.1 spec is the contract. Seven first-class shapes: observations, profiles, samples, events, collections, annotations, boundaries.
Auth
Sign in, sessions, accounts, ORCID linking
Magic-link first, optional email+password. 15-min JWT in HttpOnly cookie + 30-day rotated refresh. ORCID linking for the researcher tier. UI at mud2dust.com/auth/sign-in/.
Scientific plan
Vision, architecture, phased build, funding, governance
Lives in the repo as docs/plan/scientific-plan.md so it evolves with the work. The “what's done” section is auto-generated from BUILD-PROGRESS.md.
Status — public + auth live, multi-source ingest
As of 2026-05-08
- Map. 19 layers live at mud2dust.com: VWC v0 fusion (headline), OPERA σ⁰, HLS NDVI/NDWI/NDRE/LST, SMAP L4, MRMS / IMERG / HRRR precip, GOES-18 LST, OpenET monthly ET, Copernicus DEM, SoilGrids clay/sand/SOC/pH, USDA CDL crop type, OPERA static LIA.
- Tile API.
https://mud2dust.io/cog/tiles/…on titiler + CloudFront. - Sensor / partner API.
https://api.mud2dust.io/v1/…on Hono + Lambda + Aurora + Timescale Cloud. POST /v1/stations, /v1/observations with QC + coordinate fuzzing + per-station trust weighting (P4.20). - Auth. Magic-link + email/password sign-in, account self-serve at mud2dust.com/auth/sign-in/. JWT + rotated refresh, ORCID linking flips researcher tier. (P3.6 + P3.7 contributor dashboard.)
- Pull adapters. NRCS SCAN/SNOTEL (P4.10), AmeriFlux v2 (P4.9) — both live against real partner APIs.
What's next
- Partner-farm anchor stations (in-field soil moisture + weather probes from real working farms) → unblocks the calibration RF training (P2.6 iter 2). Headline scientific deliverable. Hardware is vendor-agnostic; the bridge ingests from any logger or vendor cloud that can hand off observations.
- L-Band drone cal/val partner → P4.13 + NASA-CSDA grant story.
- Drone Event upload (P4.11/.12) — presigned S3 multipart + Fargate preprocessing.
- Annotation + Sample + Boundary shapes(P4.15/.16/.17) — schema lands when the first contributor needs them.
What this isn't
mud2dust isn't a SaaS wrapper around a single satellite source. It's a calibrated multi-source soil-moisture model with an open contribution platform underneath, designed so distributed L-Band drone cal/val campaigns over instrumented fields produce SMAPVEX-class anchor data outside NASA campaigns. The map is the shopfront; the platform is what makes it possible.