On the use of remote sensing vegetation indices for agricultural zones delineation

Hasanli, Gunay (2026) On the use of remote sensing vegetation indices for agricultural zones delineation, [Dissertation thesis], Alma Mater Studiorum Università di Bologna. Dottorato di ricerca in Scienze e tecnologie agrarie, ambientali e alimentari, 38 Ciclo. DOI 10.48676/unibo/amsdottorato/13234.
Documenti full-text disponibili:
[thumbnail of Gunay_Hasanli_thesis.pdf] Documento PDF (English) - Richiede un lettore di PDF come Xpdf o Adobe Acrobat Reader
Disponibile con Licenza: Creative Commons: Attribuzione - Non Commerciale - Non Opere Derivate 4.0 (CC BY-NC-ND 4.0) .
Download (2MB)

Abstract

The increasing pressures of climate change, resource scarcity, environmental degradation, and growing food demand have intensified the need for more efficient and sustainable agricultural management practices. Precision agriculture addresses these challenges by accounting for the spatial and temporal variability of agricultural systems. A key component of this approach is the delineation of management zones, which enables site-specific interventions and optimized resource allocation. This thesis investigates the potential of optical remote sensing vegetation indices for agricultural zone delineation and proposes a framework for converting satellite-derived information into operational decision-support tools for precision farming. The study integrates multi-temporal Sentinel-2 multispectral imagery, vegetation index analysis, unsupervised clustering techniques, and quantitative validation procedures to identify stable patterns of within-field variability. Three widely used vegetation indices—Normalized Difference Vegetation Index (NDVI), Soil-Adjusted Vegetation Index (SAVI), and Modified Soil-Adjusted Vegetation Index (MSAVI)—were evaluated for their ability to capture differences in crop vigor, canopy development, and soil background effects. The methodology was applied to two agricultural systems in the Emilia-Romagna region of northern Italy: a walnut orchard in Bondeno and a vineyard in Imola. Management zones were delineated using vegetation index time series analyzed through K-means, Gaussian Mixture Models (GMM), and Self-Organizing Maps (SOM). The resulting zonation maps were validated against independent soil-based reference datasets derived from soil texture characterization and Time Domain Reflectometry (TDR) surveys. Clustering performance and spatial agreement were assessed using the V-measure metric. Results show that vegetation index performance is strongly influenced by crop phenology and canopy conditions. While NDVI effectively captured vegetation vigor during intermediate growth stages, SAVI and MSAVI performed better under sparse vegetation and strong soil influence. Multi-temporal analyses improved zoning reliability compared with single-date observations, whereas SOM and GMM more effectively represented complex spatial variability. Overall, the proposed framework supports data-driven, resource-efficient, and sustainable precision agriculture.

Abstract
Tipologia del documento
Tesi di dottorato
Autore
Hasanli, Gunay
Supervisore
Co-supervisore
Dottorato di ricerca
Ciclo
38
Coordinatore
Settore disciplinare
Settore concorsuale
Parole chiave
Precision Agriculture, Remote Sensing, Vegetation Indices, Agricultural Zone Delineation, Sentinel-2 Imagery, Multi-Temporal Analysis, Unsupervised Clustering
DOI
10.48676/unibo/amsdottorato/13234
Data di discussione
23 Luglio 2026
URI

Altri metadati

Statistica sui download

Gestione del documento: Visualizza la tesi

^