GPS → ThingSpeak + Auto-Parity + Merged Table (AQI + Speed)

Sensor data is continously sent from ESP32 to Thingspeak. Press the 'Start GPS Upload' button to allow device tracking and the sending of device data to Thingspeak.

ThingSpeak Visualisations

Hardware Overview

This system uses the following components:

• ESP32-DEVKITC-VE
• SCD30: CO₂, Temperature, Humidity
• SPS30: PM1.0, PM2.5, PM10
• MiCS-4514: NO₂
• DS3231 RTC
• MicroSD Card Module
• LCD Display
• Li-Ion Battery Pack

Enclosure & Mounting

Enclosure

Electronics are housed in a water-resistant plastic lunchbox with two dedicated inlets positioned directly over the SPS30 and SCD30. Each inlet is covered by a small, downward-facing hood (cut from a plastic drinks bottle) to promote ambient airflow while shedding rain and spray. The enclosure exterior is wrapped in aluminium foil tape, which reduced solar heating and eliminated missed reads observed during early tests.

• Weather protection: downward hoods + enclosure lip to minimize water ingress.
• Thermal control: reflective foil tape to limit internal temperature rise.
• Layout: sensors placed near inlets; cabling strain-relieved; SD card and lithium battery accessible.

Airflow orientation

Inlets are oriented cross-wind with the hoods angled slightly downward. This reduces splash entry, and improves representativeness of ambient air over the roof. The unit is mounted away from the exhaust path of its vehicle and other vehicles on the road.

Mounting

The enclosure is secured on the vehicle roof using a high-strength magnet (from a trailer-light mount) plus two suction cups; bungee straps across the lid provide redundancy. This configuration remained stable at motorway speeds (tested up to ~120 km/h) and maintained consistent airflow to the sensor inlets.

• Primary retention: magnet base on roof centreline.
• Secondary retention: two suction cups + crossed bungee straps.

Pre-drive checks

• “Tug test” on magnet/suction fittings; verify bungee tension.
• Confirm hoods are dry/clear and not obstructed.
• Power-on self-check (Blynk/ThingSpeak live values; SD logging active).

Deployment

Sensors sampled every 5 s. During the July/August 2025 trials we logged to microSD and uploaded to ThingSpeak and Blynk. Cloud uploads used even/odd 40 s slots (sensors vs GPS), giving an effective merged cadence of ~80 s. View the results of each test on Google Maps by expanding the map below:

Test 1 — Tue 12 Aug (Laois → Dublin, one-way)

• Peak AQI & GPS: Highest AQI values (>80) near Junction 14 (busy interchange).
• Lowest AQI & GPS: Rural sections near J16 (Laois) and after exiting J13 (Kildare).
• PM vs CO₂: Peaks do not always coincide; high CO₂ can occur with low PM.
• Speed: Some elevated PM at motorway speeds (>100 km/h).
• Conditions: On-board ~24–37 °C, RH ~20–57%; several high-AQI periods coincided with RH >55% and lower temps.

Test 2 — Wed 6 Aug (Round trip)

• Overall levels: Lower PM than Test 1; baseline ~2–6 µg/m³ with peaks ~10–12 µg/m³ near Dublin.
• Hotspots: Short-lived, location-specific PM spikes at junctions/bottlenecks detected on-road.
• PM vs CO₂: CO₂ generally 150–250 ppm with peaks at traffic bottlenecks; PM and CO₂ peaks often decoupled.
• Pairing: Sensor–GPS merge by nearest timestamp (≤90 s); points shown are single, unaveraged measurements.

Test 3 — Wed 13 Aug (Round trip, same weekday one week later)

• Repeatability: PM₂.₅ profile closely matches Test 2 (baseline ~2–6 µg/m³; peaks ~10–12 µg/m³ near Dublin).
• Hotspots: Similar spatial peaks to Test 2, reinforcing junction/bottleneck effects.
• PM vs CO₂: CO₂ behaviour similar to Test 2; moderate rises at congestion, not always aligned with PM.
• Conditions: On-board ~21–36 °C, RH ~32–67%; consistent with late-afternoon/early-evening runs.

Note: Some nearby fixed stations were inactive on our dates (e.g., Ballymun Library, Emo), so fixed–mobile comparisons were performed only where fixed data were available.