Sozzi et al. (2026) Assessing vineyard irrigation uniformity and drip system malfunction by remote and ground sensing: Insights from Sentinel-1, Sentinel-2 and Planet monitoring

Identification

• Journal: Computers and Electronics in Agriculture
• Year: 2026
• Date: 2026-04-04
• Authors: Marco Sozzi, Francesco Salmaso, Alessia Cogato, Lucia Bortolini
• DOI: 10.1016/j.compag.2026.111722

Research Groups

• Department of Land Environment Agriculture and Forestry, University of Padova, Legnaro, Italy

Short Summary

This study evaluates the effectiveness of Sentinel-1, Sentinel-2, and Planet satellite data for monitoring drip irrigation uniformity and detecting malfunctions in a vineyard, demonstrating their potential to optimize water management in viticulture.

Objective

• To identify efficient and effective satellite-based methods for assessing drip irrigation uniformity and detecting malfunctions in a vineyard, evaluating their performance before and after corrective maintenance.

Study Configuration

• Spatial Scale: A 6200 m² commercial vineyard plot (11 rows, each 225 m long) in Monselice, Northeastern Italy.
• Temporal Scale: The 2019 growing season, with core analysis from June to August 2019. Satellite data for Sentinel-1 covered January 1st to December 31st, 2019.

Methodology and Data

• Models used:
  • Volumetric Water Content (VWC) estimation from Sentinel-1 using a change-detection approach.
  • Normalized Difference Vegetation Index (NDVI) calculation.
  • Christiansen Uniformity Coefficient (CU) and Low-Quarter Distribution Uniformity (DULQ) for ground truth.
  • Satellite-Derived Uniformity Coefficient (SDCU) and Satellite-Derived Low-Quarter Distribution Uniformity (SDDULQ).
  • Ordinary kriging for interpolating ground measurements.
  • Lee filter for Sentinel-1 speckle noise reduction.
  • Incidence angle normalization for Sentinel-1.
  • Cloud masking for Sentinel-2.
  • Linear regression models and Pearson correlation coefficient (r) for statistical analysis.
  • Type II ANOVA for sensitivity analysis of Sentinel-1 VWC.
• Data sources:
  • Satellite:
    • Sentinel-1: C-band Synthetic Aperture Radar (SAR) (VV polarization), 10 m spatial resolution, effective temporal resolution of approximately 2 days.
    • Sentinel-2: Multispectral Instrument (MSI) (B4 red, B8 near-infrared), 10 m spatial resolution, effective temporal resolution of approximately 9 days.
    • PlanetScope: Multispectral (blue, green, red, near-infrared), 3-5 m spatial resolution, effective temporal resolution of approximately 3 days.
  • Ground observation:
    • Catch-can measurements of emitter discharge rate at 16 points (4 cans per point) following ISO 9261 (2004) protocol.
    • Field inspections for identifying drip system malfunctions.
  • Climatic data: Weather station located 3 km from the study area (ARPAV, 2019).

Main Results

• Initial ground measurements (June 12th) showed suboptimal irrigation uniformity (DULQ = 0.57, CU = 74.9%).
• After maintenance (July 23rd), uniformity significantly improved (DULQ = 0.81, CU = 92.3%).
• Sentinel-1 VWC exhibited the lowest bias for satellite-derived uniformity indices compared to ground truth: SDCU bias of 7.9% and SDDULQ bias of 33.8%.
• Sentinel-2 and Planet NDVI showed higher biases: SDCU biases of 15.2% and 16.8%, and SDDULQ biases of 40.9% and 41.9%, respectively.
• Sentinel-1 VWC correlated significantly with catch-can discharge rate on 19 out of 45 days (42.2% of observations), with Pearson correlation coefficients (r) ranging from -0.818 to 0.909 (R-squared up to 0.83). It allowed for immediate soil moisture anomaly detection post-irrigation.
• Sentinel-2 NDVI correlated significantly on 4 out of 10 days (40.0% of observations), with r ranging from 0.568 to 0.704 (R-squared up to 0.50).
• Planet NDVI correlated significantly on 3 out of 30 days (10.0% of observations), with r ranging from 0.503 to 0.681 (R-squared up to 0.46).
• NDVI from optical sensors effectively captured delayed vegetative responses, typically around 15 days after an irrigation event, reflecting long-term vineyard health.
• The higher bias for SDDULQ was attributed to its sensitivity to localized low-discharge extremes, which are smoothed by the spatial averaging of satellite pixels (3-10 m resolution).
• A Type II ANOVA revealed no significant influence of Sentinel-1 acquisition conditions (sensors, acquisition directions, orbital geometries) on VWC variations.

Contributions

• Demonstrates the complementary roles of SAR (Sentinel-1 VWC for immediate soil moisture response) and optical (Sentinel-2 and Planet NDVI for delayed vegetative response) satellite data in assessing drip irrigation uniformity and detecting malfunctions at the farm level.
• Introduces Satellite-Derived Uniformity Coefficient (SDCU) and Satellite-Derived Low-Quarter Distribution Uniformity (SDDULQ) as efficient, scalable alternatives to traditional manual methods for irrigation assessment.
• Provides quantitative insights into the accuracy and bias of satellite-derived uniformity indices, highlighting Sentinel-1 VWC's superior performance for immediate detection.
• Advocates for the broader application of integrated satellite data to support sustainable viticulture practices, particularly in regions facing variable climatic conditions and water scarcity.
• Contributes to optimizing "blue water" use by identifying system malfunctions and minimizing unproductive water consumption.

Funding

Not explicitly mentioned in the provided text.

Citation

@article{Sozzi2026Assessing,
  author = {Sozzi, Marco and Salmaso, Francesco and Cogato, Alessia and Bortolini, Lucia},
  title = {Assessing vineyard irrigation uniformity and drip system malfunction by remote and ground sensing: Insights from Sentinel-1, Sentinel-2 and Planet monitoring},
  journal = {Computers and Electronics in Agriculture},
  year = {2026},
  doi = {10.1016/j.compag.2026.111722},
  url = {https://doi.org/10.1016/j.compag.2026.111722}
}