Abstract

Reliable, site-specific hydrometeorological observations are essential for supporting eco-hydrological analysis and water resources management. This study developed and evaluated a low-cost Internet of Things (IoT)-based Mini-Automatic Weather Station (Mini-AWS) integrating an ESP-12E/F microcontroller, temperature and humidity sensors, a tipping-bucket rain gauge, local data storage, wireless communication, and a solar-assisted battery system. Field evaluation was conducted during August 2026 using observations from the Indonesian Meteorology, Climatology, and Geophysics Agency (BMKG). Analysis of 43,967 synchronized temperature and relative humidity pairs yielded a temperature RMSE of 1.066 °C and MAPE of 2.550%. The uncalibrated humidity measurements exhibited an RMSE of 10.881 %RH and a positive measurement bias. A second-order polynomial humidity calibration model was selected using temporally separated training and validation blocks and evaluated on 10,052 internal test observations. Calibration reduced humidity RMSE from 11.500 to 2.848 %RH and MAPE from 17.165% to 3.938%, although residual autocorrelation and heteroscedasticity remained. The correction also reduced the RMSE of AWS-derived Vapor Pressure Deficit (VPD) from 0.3701 to 0.1668 kPa and improved agreement with BMKG-derived operational VPD categories from 45.09% to 81.86%. Initial volumetric calibration of the tipping bucket yielded 0.5084 mm/tip, while 15 additional controlled-flow experiments produced a mean resolution of 0.544 mm/tip; natural-rainfall accuracy was not independently validated. The prototype acquired meteorological data during a 31-day deployment using a solar-assisted 2500 mAh Li-ion battery, although battery-voltage accuracy and energy savings were not independently verified. The dashboard incorporated VPD, empirical Ivanov Potential Evapotranspiration (PET), and Rain-Free Days indicators as preliminary eco-hydrological interpretation tools. The calibration and VPD improvements were evaluated offline, while physical implementation of the polynomial correction on the microcontroller remains to be verified. Overall, the Mini-AWS demonstrated promising monitoring performance and improved humidity-derived indicators, subject to further operational and cross-season validation.