Ye et al. (2026) An IoT-based predictive irrigation scheduling framework for precision soil moisture control in greenhouses

Identification

• Journal: Agricultural Water Management
• Year: 2026
• Date: 2026-04-10
• Authors: Xujun Ye, Natsumi Shirakawa, Kaisyu Sano, Shuhuai Zhang
• DOI: 10.1016/j.agwat.2026.110334

Research Groups

• Faculty of Agriculture and Life Science, Hirosaki University, Japan
• Graduate School of Agriculture and Life Science, Hirosaki University, Japan

Short Summary

This study developed an IoT-based predictive irrigation framework that optimizes soil moisture control by establishing a proportional relationship (Δtpred = ½Δteval) between evaluation and prediction intervals, demonstrating stable moisture regulation and reduced irrigation frequency in greenhouse experiments.

Objective

• To develop and validate an IoT-based predictive irrigation scheduling framework that coordinates evaluation (Δteval) and prediction (Δtpred) intervals for precision soil moisture control, with an emphasis on calibrating their quantitative relationship to ensure stable and responsive regulation.

Study Configuration

• Spatial Scale: Pot cultivation (4 L soil volume per pot) in a semi-controlled greenhouse environment. Experiments involved bare soil, cabbage (Brassica oleracea var. capitata), and tomato (Solanum lycopersicum L.) plants. Soil moisture sensors were inserted at a depth of 5–7 cm in each pot.
• Temporal Scale: Soil water content data were collected every 5 minutes. The first experiment (bare soil and cabbage) ran from May 9 to June 13, 2025. The second experiment (tomato) ran from July 2 to August 31, 2025. Evaluation intervals (Δteval) tested ranged from 30 minutes to 180 minutes, with corresponding prediction intervals (Δtpred) from 15 minutes to 90 minutes.

Methodology and Data

• Models used:
  • Theoretical analysis of the relationship between Δteval and Δtpred.
  • Numerical simulations using a linear model for soil water content depletion.
  • A third-order polynomial calibration model (R²=0.999) to convert TEROS 12 sensor readings to volumetric water content (VWC).
• Data sources:
  • Real-time soil moisture data from TEROS 12 sensors (METER Group AG).
  • Irrigation volumes delivered by Unit Watering U101 actuators (M5STACK).
  • System control and data logging managed by M5StickCPlus microcontrollers (M5STACK) and transmitted to Google Spreadsheet.
  • Meteorological data (air temperature, relative humidity) from the Japan Meteorological Agency.

Main Results

• Theoretical analysis and simulations demonstrated that setting the prediction interval to half of the evaluation interval (Δtpred = ½Δteval) consistently maintains the average soil water content precisely at the target threshold, irrespective of soil water depletion rates.
• Greenhouse experiments with bare soil and cabbage successfully regulated soil moisture near the target of 29% VWC (2350 a.u.). Shorter evaluation intervals improved control precision but increased irrigation frequency, while longer intervals maintained comparable target levels with fewer activations, indicating improved energy efficiency. The maximum observed deviation from the target was 0.6% of the threshold.
• The irrigation system accurately quantified plant water consumption, with an estimated total of 1.2123 L for cabbage over its experimental period.
• Greenhouse experiments with tomato plants successfully maintained distinct sequential VWC target levels ranging from 10% to 30% across multiple pots, demonstrating robust and stable regulation even under varying environmental conditions.
• Discrepancies between simulations and experiments were noted due to physical constraints like infiltration delay, which caused a lag of approximately 5–20 minutes between water application and sensor response.

Contributions

• Introduced a novel temporal optimization strategy for IoT-based irrigation by establishing a quantitative proportional relationship (Δtpred = ½Δteval) between evaluation and prediction intervals.
• Resolved a common structural mismatch in IoT-based irrigation systems, leading to stable soil moisture regulation and reduced oscillatory behavior around target levels.
• Provided a model-agnostic framework that can be integrated with various prediction models (e.g., linear, evapotranspiration-based, machine learning) to enhance system stability and interpretability.
• Demonstrated potential for improved energy efficiency and operational sustainability by reducing pump activation frequency and mechanical stress on irrigation components.
• Enabled accurate quantification of plant water consumption, enhancing precision in crop water use estimation for greenhouse production systems.

Funding

• Japan Society for the Promotion of Science (JSPS) under the Grants-in-Aid for Scientific Research (C), Grant Number 23K05459.
• JSPS Program for Forming Japan’s Peak Research Universities (J-PEAKS), Grant Number JPJS00420240013.

Citation

@article{Ye2026IoTbased,
  author = {Ye, Xujun and Shirakawa, Natsumi and Sano, Kaisyu and Zhang, Shuhuai},
  title = {An IoT-based predictive irrigation scheduling framework for precision soil moisture control in greenhouses},
  journal = {Agricultural Water Management},
  year = {2026},
  doi = {10.1016/j.agwat.2026.110334},
  url = {https://doi.org/10.1016/j.agwat.2026.110334}
}