Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Mapping of Weed Management Methods in Orchards using Sentinel-2 and PlanetScope Data

Ioannis Kontogiorgakis, Iason Tsardanidis, Dimitrios Bormpoudakis, Ilias Tsoumas, Dimitra A. Loka, Christos Noulas, Alexandros Tsitouras, Charalampos Kontoes

TL;DR

The paper tackles the challenge of monitoring weed management practices in orchards by leveraging time-series data from Sentinel-2 and PlanetScope and applying machine learning to classify four practices: Mowing, Tillage, Chemical Spraying, and No practice. It builds a parcel-based, time-series classification framework that fuses multi-sensor spectral features, including NDVI dynamics, and evaluates RF, XGBoost, and KNN. Results show PlanetScope data yields higher accuracy than Sentinel-2, with Random Forest on PS achieving a weighted F1 of about 0.57, and mowing being the most detectable practice while chemical spraying and no practice remain difficult due to subtle spectral signals and limited data. The work demonstrates the potential of EO-driven methods for scalable weed-management mapping in orchards and points to data enrichment and sensor fusion as avenues to improve performance for policy and agro-ecosystem monitoring.

Abstract

Effective weed management is crucial for improving agricultural productivity, as weeds compete with crops for vital resources like nutrients and water. Accurate maps of weed management methods are essential for policymakers to assess farmer practices, evaluate impacts on vegetation health, biodiversity, and climate, as well as ensure compliance with policies and subsidies. However, monitoring weed management methods is challenging as they commonly rely on ground-based field surveys, which are often costly, time-consuming and subject to delays. In order to tackle this problem, we leverage earth observation data and Machine Learning (ML). Specifically, we developed separate ML models using Sentinel-2 and PlanetScope satellite time series data, respectively, to classify four distinct weed management methods (Mowing, Tillage, Chemical-spraying, and No practice) in orchards. The findings demonstrate the potential of ML-driven remote sensing to enhance the efficiency and accuracy of weed management mapping in orchards.

Mapping of Weed Management Methods in Orchards using Sentinel-2 and PlanetScope Data

TL;DR

The paper tackles the challenge of monitoring weed management practices in orchards by leveraging time-series data from Sentinel-2 and PlanetScope and applying machine learning to classify four practices: Mowing, Tillage, Chemical Spraying, and No practice. It builds a parcel-based, time-series classification framework that fuses multi-sensor spectral features, including NDVI dynamics, and evaluates RF, XGBoost, and KNN. Results show PlanetScope data yields higher accuracy than Sentinel-2, with Random Forest on PS achieving a weighted F1 of about 0.57, and mowing being the most detectable practice while chemical spraying and no practice remain difficult due to subtle spectral signals and limited data. The work demonstrates the potential of EO-driven methods for scalable weed-management mapping in orchards and points to data enrichment and sensor fusion as avenues to improve performance for policy and agro-ecosystem monitoring.

Abstract

Effective weed management is crucial for improving agricultural productivity, as weeds compete with crops for vital resources like nutrients and water. Accurate maps of weed management methods are essential for policymakers to assess farmer practices, evaluate impacts on vegetation health, biodiversity, and climate, as well as ensure compliance with policies and subsidies. However, monitoring weed management methods is challenging as they commonly rely on ground-based field surveys, which are often costly, time-consuming and subject to delays. In order to tackle this problem, we leverage earth observation data and Machine Learning (ML). Specifically, we developed separate ML models using Sentinel-2 and PlanetScope satellite time series data, respectively, to classify four distinct weed management methods (Mowing, Tillage, Chemical-spraying, and No practice) in orchards. The findings demonstrate the potential of ML-driven remote sensing to enhance the efficiency and accuracy of weed management mapping in orchards.
Paper Structure (11 sections, 4 equations, 3 figures, 3 tables)

This paper contains 11 sections, 4 equations, 3 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Data Collection
  3. Point Dataset
  4. Satellite Dataset
  5. Feature Extraction
  6. Methodology
  7. Classification approach
  8. Experimental Setup
  9. Results and Discussion
  10. Conclusion & Future Work
  11. Acknowledgements

Figures (3)

  • Figure 1: Our area of interest in Thessaly, Greece. Yellow points represent the fields where we identified a weed management method.
  • Figure 2: Overview of the methodology pipeline for weed management method classification. The process includes data collection, data pre-processing, exploratory data analysis, feature engineering, training a classification model and evaluating its performance.
  • Figure 3: Confusion matrix for the best RF model trained and evaluated on PS data. Warmer colors indicate higher agreement (i.e., more correct predictions), while colder colors represent lower agreement.