Boussetta et al. (2026) Long-term NDVI series analysis in a sylvopastoral irrigated area

Identification

• Journal: The Egyptian Journal of Remote Sensing and Space Science
• Year: 2026
• Date: 2026-03-31
• Authors: Inès Boussetta, Nesrine Abid, Ahmed Ezzine, Zoubeïda Bargaoui
• DOI: 10.1016/j.ejrs.2026.03.002

Research Groups

• National School of Engineers of Tunis (ENIT), University Tunis El Manar, Civil Engineering Department, Water and Environment Research Laboratory, ENIT, Tunis, Tunisia
• Centre National de Cartographie et Télédétection (CNCT), Tunis, Tunisia

Short Summary

This study analyzes long-term Normalized Difference Vegetation Index (NDVI) series from 1984 to 2017 in a sylvopastoral irrigated area in northern Tunisia to assess vegetation productivity changes. It reveals a significant increase in crop productivity from 2010, primarily driven by increased irrigation inputs and phenological growth, particularly during rainfall shortage years, where irrigation volumes strongly predict NDVI.

Objective

• To identify long-term NDVI changes and vegetation productivity in a sylvopastoral irrigated area in northern Tunisia.
• To analyze these changes in relation to the introduction of irrigation facilities, rainfall, and irrigation inputs using non-stationary frequency analysis.

Study Configuration

• Spatial Scale: A 25 hectare (ha) specific area of interest within a 120 ha public irrigation district in Bargou Mountain, northern Tunisia, analyzed at 30 meter (m) spatial resolution.
• Temporal Scale: A 34-year period from 1984 to 2017, with annual June NDVI observations.

Methodology and Data

• Models used:
  • Normalized Difference Vegetation Index (NDVI) calculation.
  • Statistical fitting of Normal and two-parameter Log-Normal distributions (Kolmogorov-Smirnov test).
  • Moving average for trend analysis.
  • Linear regression for analyzing parameter dependencies.
  • Logarithmic regression for predictive modeling.
• Data sources:
  • Satellite: Landsat (Thematic Mapper (TM) and Operational Land Imager (OLI)) images (30 m resolution, June, 1984-2017, USGS Level 2). Sentinel-2A images for validation.
  • Observation: Field interviews with farmers from 6 orchards (21 pixels), daily rainfall data from Saadia Bargou station (1983-2017), and groundwater pumping volumes from Ministry of Agriculture yearbooks (1984-2017).

Main Results

• Landsat NDVI estimations showed acceptable concordance (95% and 75%) with Sentinel-2A data for two dates, despite a slight underestimation.
• Field interviews validated the coherence between NDVI categories and reported land use changes and crop development stages over time.
• Both the median NDVI and the interquartile range of NDVI maps exhibited upward trends, indicating increased crop productivity and dynamic landscape changes.
• The two-parameter Log-Normal distribution adequately represented NDVI spatial variability for 76% of the years, consistently from 1998 onwards.
• The Log-Normal location parameter (μ) showed an upward trend, while the scale parameter (σ) showed a plateau followed by an increase from 2010.
• A scatterplot of standardized μ and σ² identified two distinct periods (C1: 1998-2010 and C2: 2011-2017), marking 2010 as a significant year of change in vegetation productivity.
• Linear regression analysis revealed that the location parameter μ increased significantly after 2010 for the same growing season rainfall conditions, attributed to phenological growth and increased irrigation inputs.
• A significant predictive logarithmic model (coefficient of determination R² = 0.62) was established between median NDVI and irrigation volumes per unit area for years with rainfall shortage, outperforming models using total water inputs.
• During precipitation shortage years, the average irrigation input was 1444 millimeter per hectare (mm/ha) per NDVI unit, compared to 1269 mm/ha for surplus years, suggesting potential for water saving during surplus seasons.
• Average irrigation inputs in the study area (2002 cubic meters per hectare per year (m³/ha/year) or 5.5 mm/day) are consistent with high values reported for apple tree orchards globally.

Contributions

• Provides the first long-term (1984-2017) NDVI analysis for this specific sylvopastoral irrigated area in Tunisia, linking vegetation productivity directly to irrigation and rainfall.
• Innovatively applies non-stationary frequency analysis using Log-Normal distributions to NDVI time series, interpreting the evolution of its parameters (location and scale) in relation to anthropogenic and climatic factors.
• Identifies a critical change point (2010) in vegetation productivity, quantitatively linking it to increased irrigation and phenological development of fruit trees.
• Develops a robust predictive logarithmic model for median NDVI based on irrigation volumes, demonstrating its particular effectiveness and importance for water management during rainfall deficit years.
• Quantifies the benefits of irrigation during rainfall deficit periods and highlights opportunities for optimizing water use efficiency during surplus rainfall conditions.

Funding

• PRIMA project “SWATCH: Strategies for increasing the water use efficiency of semi-arid Mediterranean watersheds and agrosilvopastoral systems under climate change” (https://swatchprima.com/)
• Tunisian Ministry of Higher Education and Scientific Research

Citation

@article{Boussetta2026Longterm,
  author = {Boussetta, Inès and Abid, Nesrine and Ezzine, Ahmed and Bargaoui, Zoubeïda},
  title = {Long-term NDVI series analysis in a sylvopastoral irrigated area},
  journal = {The Egyptian Journal of Remote Sensing and Space Science},
  year = {2026},
  doi = {10.1016/j.ejrs.2026.03.002},
  url = {https://doi.org/10.1016/j.ejrs.2026.03.002}
}