ESP32 water quality monitor — 7 sensors + GPS, Firebase upload every 5s.
Water quality monitoring in Indonesian waterways is predominantly manual — researchers collect grab samples, transport them to a lab, and wait days for results. By the time data arrives, conditions have changed and the snapshot is already stale. Continuous real-time monitoring remains rare outside major cities.
Commercial water quality sondes with multi-parameter capability cost $3,000–$15,000, putting them out of reach for universities, local environmental agencies, and community water monitoring groups. Even when equipment is available, spatial coverage is limited to one or two fixed stations.
There was a clear gap for a low-cost, portable, multi-sensor system that could measure 7 water parameters simultaneously, geotag every reading with GPS, and upload data to the cloud in near-real-time for accessible dashboard visualization.
The system integrates seven water quality sensors with an ESP32-WROOM-32 microcontroller. Sensors connect via analog inputs and I2C interfaces. The ESP32 reads all 7 channels simultaneously, packages them into a JSON payload with GPS coordinates from the NEO-6M module, and uploads to Firebase Realtime Database every 5 seconds over Wi-Fi.
Sigit designed the sensor interface board with signal conditioning for each probe type — pH and ORP require high-impedance buffers, DO needs temperature compensation, and turbidity uses a photodiode amplifier. Power comes from two 18650 Li-Ion cells with a solar panel for extended field deployment.
The GPS geotagging enables spatial water quality mapping — as the operator moves along a river or lake, each reading is stamped with coordinates, building a continuous spatial profile. Firebase provides real-time synchronization to a web dashboard where multiple stakeholders can monitor conditions simultaneously.
The 5-second sampling rate captures transient events — pollution pulses, algal bloom onset, or stratification changes — that hourly or daily sampling would miss entirely. Data is stored in Firebase for historical analysis and CSV export.
The ArsenikV2 system was field-tested across six water bodies in East Java: Kali Mas, Kali Surabaya, Kali Wonokromo, Kali Brantas, Kali Lamong, and Kali Porong. Measurements were compared against a calibrated Horiba U-50 multiparameter water quality meter as reference. All seven sensors achieved error rates under 5% of the reference instrument across all six sites.
The 5-second sampling rate generated over 17,000 data points per sensor per day, capturing diurnal patterns invisible to grab sampling. GPS-tagged data revealed spatial gradients — dissolved oxygen declining downstream from Kali Brantas toward Kali Porong, turbidity spiking near urban runoff points, and conductivity increasing near industrial discharge zones.
Firebase dashboard latency averaged 1.8 seconds from sensor read to dashboard display, well within the target for near-real-time monitoring. The solar-powered configuration sustained continuous operation for 48+ hours without external charging, sufficient for multi-day field campaigns.
Sensor calibration followed a two-point protocol per parameter. pH was calibrated with buffer solutions at pH 4.0 and 7.0. Conductivity used 1413 µS/cm and 12.88 mS/cm standards. Dissolved oxygen was calibrated against water-saturated air (100% saturation method). Turbidity used 0 NTU (distilled water) and 100 NTU formazin standards. ORP used Zobell's solution at 228 mV. TDS was derived from conductivity with a 0.5 conversion factor calibrated against gravimetric measurements.
The field protocol involved a 2-minute sensor stabilization period at each measurement station before logging began. Measurements continued for a 5-minute window at each site, producing 60 readings per sensor per site. Outlier rejection used the median absolute deviation (MAD) method with a 3σ threshold. Reported values are the trimmed mean of the remaining measurements.
Validation against the Horiba U-50 reference instrument was performed at each of the six sites. The reference instrument was calibrated daily before field deployment. Comparison readings were taken within 30 seconds of each other to minimize temporal variation. Error was calculated as the percentage deviation from the reference for each parameter at each site, then averaged across sites for the reported accuracy figures.
International Journal of Engineering Research and Technology (IJERT), 2023
Water quality instrumentation research
Shares Firebase real-time upload pattern and ESP32 sensor integration for precision agriculture water monitoring.
Uses the same ESP32 + Firebase + GPS architecture for environmental monitoring with multi-sensor array.
Extends the low-cost sensor instrumentation approach to water quality measurement with GPS spatial analysis.
Solar resource methodology used for sizing the autonomous power system of field-deployed water quality monitors.
This instrumentation work extends sigit's embedded systems expertise to environmental water quality monitoring, connecting the sensor array, Firebase IoT pipeline, and GPS spatial analysis capabilities demonstrated across the portfolio.