README.md

---
pretty_name: STAF data (browser mirror)
tags:
- hydrology
- streams
- nhdplus
- geospatial
---

# STAF data: the browser mirror

The data the static STAF web app (EASI, SFARI and DEEP; the Stream Tiered Assessment Framework,
https://usace-wrises.github.io/staf/) reads by HTTP range. Every file mirrors an asset of a GitHub
release of https://github.com/USACE-WRISES/staf, which stays the source:

- `bundle/`: the STAF data bundle (`staf-data-current`), unzipped: NHDPlus HR flowlines, catchments
  and their shared borders, precomputed per-catchment values, the NHDPlus V2 crosswalk with
  StreamCat, and national point tables (WQP, NID, NWIS, NAS), for the lower 48.
- `xs/`: the national cross-section grid (`staf-xs-current`, published region by region): every
  section's scoring numbers, what drawing or pulling a section needs, one profile per segment.
- `points/`: the bundle's national point tables (dams, stream gages, water-quality stations and
  results) sorted by place in small row groups, so a lookup near a site reads a few kilobytes.
- `library/`: the STAF assessment library (`library`): DEEP's regional assessments, calculators and
  EASI's method packages.
- `engine/`: the data files the STAF site engine ships with: the USGS base-flow index grid (Wolock
  2003), the EPA Level III ecoregions and the physiographic divisions.

The files the app reads a few rows of at a time (each region's stream lines, StreamCat rows, V2
attributes and point extras, and the cross-section files) are written again with the same rows and values
in small pages with a page index, so a read takes kilobytes; `topo/` holds each region's network topology
on its own. JSON and GeoJSON files are stored gzipped (`.gz`).

Files are stored under content-hashed names and never change in place; `web-manifest.json` (updated
last) maps each logical name to its file, size and sha256. Built from public federal datasets (USGS,
EPA, USACE, NOAA, USDA, FWS); see the STAF repository for sources and methods.