A LoRaWAN-Enabled Capacitive Sensor Network for Soil Moisture and Groundwater Monitoring

Water scarcity requires low-cost systems capable of continuously monitoring soil water and groundwater dynamics in agricultural fields. We developed a LoRaWAN-enabled sensing platform integrating a wire-based interdigital capacitive soil-moisture probe fabricated on a PMMA substrate and a U-shaped PTFE-insulated capacitive groundwater probe installed in a perforated PVC tube. Both sensors are interfaced with an ESP32 and communicate through 868 MHz LoRaWAN, with cloud-side calibration for soil moisture. Soil-moisture calibration against the gravimetric reference at 15, 30, and 45 cm yielded R² values of 0.975-0.988 and RMSE values of 1.08-1.66% v/v. Freshwater groundwater calibration over the measured 0-200 cm range yielded R² = 0.99946 and RMSE = 1.35 cm. The salinity experiment comprised 990 raw observations from 12 sensor probes and 12 solution batches. Outcome-independent robust QC flagged 60 observations (6.06%), leaving 930 observations for analysis. The pooled quadratic calibration achieved R² = 0.99771 and RMSE = 3.02 cm, while EC-specific models yielded R² values of 0.99852-0.99892 and RMSE values of 2.08-2.45 cm. Repeated-measures analysis showed significant main effects of depth and EC but no statistically significant Depth × EC interaction. The maximum mean relative ADC deviation from the EC = 0.5 dS m^-1 reference was 1.82% at EC = 4.0 dS m^-1 and 200 cm. Field validation quantified linear association, bias, and agreement with reference instruments. Communication exceeded 90% packet delivery at 8 km in open terrain and 5 km in dense sugarcane under the tested conditions. Routine scheduled sensing required 16.60 mWh day^-1 with BLE disabled.