SPECTRA

1. Enrol and choose a password: spectra-enrol.html (already enrolled? check your status there).
2. Sign in to the SPECTRA app with your enrolment email and password: portal.placealert.org/apex/
3. Work through the lessons on this page.
4. Take the final assessment: spectra-assessment.html. Score 80% or more and your certificate is issued straight away.

About this course

SPECTRA is the trajectory‑and‑exposure web app on the Place Alert Labs platform (the PALs Portal, from placealert.org → portal.placealert.org). It turns raw GPS tracks from wearables, phones, or low‑cost sensors into personal air‑pollution & heat exposure estimates by working out, point by point, where a person was, whether they were indoors or outdoors, how long they stayed, and how much they were exposed to there.

This course takes a non‑technical user from a raw CSV to a finished, publishable analysis, using only the buttons that exist in the live app. It mirrors SPECTRA's eight modules exactly.

Course map

Sample datasets — Zimbabwe

Synthetic GPS + exposure data for the exercises (fabricated for teaching, not real measurements). Column names auto‑map in SPECTRA. Built‑in story: indoor biomass cooking → PM2.5 spikes; commuting → NO₂; midday outdoors → WBGT heat stress.

Getting an account & signing in

1. Open SPECTRA at portal.placealert.org/apex/.
2. On the dark login page click Sign up, enter email + password.
3. Confirm via the verification email (check spam; ask a PALS admin if none arrives).
4. Log in → you land on the SPECTRA home menu: seven module cards + About.

What SPECTRA actually does

CSV/GPS upload
  → Field mapping (lat? lon? person? time? exposure?)
  → Boundary filter (keep points in your study area)
  → Dwell time (how long at each point; remove "teleports")
  → Building footprints (OSM / Microsoft / Overture / custom)
  → Indoor / Outdoor classification (point-in-building + buffer + speed)
  → Micro-environment segmentation (indoor / outdoor / in-transport)
  → Grid aggregation (e.g. 100 m cells)
  → Daily summaries + dashboard + maps + quality metrics → Export

Key idea — time‑weighted exposure

SPECTRA weights each reading by how long the person dwelled there, so 8 hours indoors counts far more than a 2‑minute walk past a busy road. That is the “Personal Exposure” in the name.

Preparing your data

SPECTRA auto‑detects columns, but a clean file maps first time. One row per GPS point:

Rules that save time: decimal‑degree coordinates; one consistent timestamp format; a real Person ID column (the engine batches per person); up to ~1 GB per file. Also accepted: .rds, .geojson, .gpx (wearables), or a direct CSV / published Google‑Sheet URL.

Load data & map fields · Input & Configuration

1. From home click the Input & Configuration card.
2. Section 1 · Data Input — pick the File Format toggle (CSV / RDS / GeoJSON / GPX / URL) and upload, or paste a URL and Load from URL.
3. Section 2 · Field Mapping — confirm/fix the auto‑filled guesses: Coordinates (Lat, Lon), Identification (Person ID, Date/Time), Exposure Variables (map your pollutant to Exposure 1; + Add Exposure Variable for more), and the optional Ground‑truth.

Configure & run

Building source (for indoor/outdoor)

Click Run. Large datasets process in per‑person batches; a progress indicator shows the stage, and the live map previews your points.

Reading the Dashboard

The Dashboard summarises the run: total points, participants, days, indoor‑vs‑outdoor share, and time‑weighted mean exposure per pollutant, with plotly charts. This is your headline result — note how indoor cooking dominates the PM2.5 dose.

Maps & trajectories · Geospatial & Tracks

Interactive Leaflet maps with switchable basemaps show classified points (indoor/outdoor), trajectory lines per person‑day, and grid cells shaded by exposure. Use them to sanity‑check classification and spot hotspots.

Quality Assessment

With a ground‑truth label, this module shows a confusion matrix plus accuracy, precision, recall, F1 for the indoor/outdoor classifier — use it to defend your method. No ground truth → informational only.

Segmentation (micro‑environments)

Breaks each track into themes — typically indoor, outdoor, in‑transport — using building location, dwell time and speed. This is what lets you say “most of this person’s PM2.5 dose came from indoor cooking, not commuting.”

Exporting & saving sessions · Data & Export

Download classified points / grid cells / daily summaries as CSV, GeoJSON, or JSON, and a Session RDS snapshot of the whole analysis.

Troubleshooting & good practice

Capstone project

Take one participant‑week from zim_spectra_master.csv and produce: (1) time‑weighted mean exposure per pollutant, (2) a hotspot map, (3) an indoor‑vs‑transport dose split, (4) a quality figure. Export everything plus a Session RDS and write a one‑paragraph interpretation — that is a CHEAQI‑MNCH mini‑analysis.

Final assessment and certificate

When you have finished the lessons and the capstone, take the final assessment: 15 multiple-choice questions, pass mark 80%, up to 3 attempts in 24 hours. Passing issues your Place Alert Labs certificate straight away, with a QR code that anyone can scan to verify it.

Take the final assessment →

Quick reference

• Always start at Input & Configuration → upload → map fields → run.
• Five required fields: Lat, Lon, Person ID, Date/Time, ≥1 Exposure.
• Faster & more accurate indoor/outdoor: use a local building footprint file.
• Reproducibility: export the Session RDS every time.
• Modules in order: Input → Dashboard → Geospatial → Quality → Segmentation → Export.