TrailPulse 🚵 ⛷️

Ever drive 45 minutes to the trailhead only to find the singletrack is a slip-n-slide or the Nordic loops are brown mush? Yeah, same. TrailPulse fixes that.

It's a Quarto website that pulls live weather, streamflow, and forecast data and mashes it into a plain-language trail conditions report for each park. Right now it covers Hunters Creek Park (Wales, NY) — 18 miles of forested singletrack.

What it shows

• Mud score — a model-based 1–10 rating of how wrecked the trails probably are, based on soil moisture, recent rain, and drainage. If it's a 9, stay home and clean your bike instead.
• Snow conditions — snow depth, recent snowfall, and whether the base is holding. Useful for deciding if it's worth waxing up.
• Current + 7-day forecast — temperature, precipitation, humidity, and streamflow plotted together so you can time your ride or ski around the weather window.
• Climatological context — shaded ribbons showing how current conditions compare to the historical average. Handy for knowing whether that 3-inch dump is actually unusual or just a Tuesday in February.

How it works

Data flows in automatically:

• Weather — Open-Meteo (historical archive + 7-day forecast)
• Streamflow — USGS gauge on the Buffalo River near Elma
• Forecast text — NWS

GitHub Actions rebuilds the site twice daily at 6:00 AM and 1:00 PM EST (UTC crons 0 11 / 0 18; during EDT that is 7:00 AM / 2:00 PM local).

Adding a park

Drop a new entry in data/parks.yml, create a parks/<park-id>/index.qmd (copy Hunters Creek as a template), tune the mud model calibration for the local soils, and you're in.

For Erie County parks, park boundaries for HLS EVI (AppEEARS) live at parks/<park-id>/data/*.geojson, referenced from boundary_file in data/parks.yml (path is relative to the park page directory). Source layer: Erie County Parks Locator (NAME / SHORT_NAME / Location / acres).

NASA Earthdata credentials (NDVI)

The NDVI panel fetches HLS Landsat + Sentinel-2 data via the NASA AppEEARS API, which requires a free NASA Earthdata account.

Local development

Add credentials to your .Renviron (run usethis::edit_r_environ() or edit ~/.Renviron directly):

EARTHDATA_USER=your_username
EARTHDATA_PASSWORD=your_password

Restart R after saving. The first render submits an async task to AppEEARS; the NDVI chart appears on the next render once the task completes (~5–20 min).

GitHub Actions

Add the same two values as repository secrets (Settings → Secrets and variables → Actions → New repository secret):

Secret name	Value
EARTHDATA_USER	your Earthdata username
EARTHDATA_PASSWORD	your Earthdata password

Then expose them as environment variables in the workflow file:

env:
  EARTHDATA_USER: ${{ secrets.EARTHDATA_USER }}
  EARTHDATA_PASSWORD: ${{ secrets.EARTHDATA_PASSWORD }}

If the credentials are missing or invalid, the NDVI panel is silently skipped — all other panels render normally.

Stack

R · Quarto · dygraphs · httr2 · renv · GitHub Actions