UAV and satellite-based prediction of aboveground biomass in scots pine stands: a comparative analysis of regression and neural network approaches

Basit Öğe Kaydı
dc.authoridAksoy, Hasan/0000-0003-1980-3834
dc.contributor.authorAksoy, Hasan
dc.contributor.authorGunlu, Alkan
dc.date.accessioned2025-03-23T19:42:27Z
dc.date.available2025-03-23T19:42:27Z
dc.date.issued2025
dc.departmentSinop Üniversitesi
dc.description.abstractForest ecosystems play a vital role in balancing the global climate through functions such as regulating carbon emissions, carbon sequestration, and energy and water cycles. Aboveground biomass (AGB) is a critical component in forest management to understand better and predict the global carbon cycle. However, traditional methods used in AGB measurement involve time-consuming, costly, and labor-intensive processes. Sentinel-1 (active), Sentinel-2, and Landsat (passive) satellite imagery, which is freely accessible and offers global coverage with frequent updates, and recently developed remote sensing platforms such as Unmanned Aerial Vehicle (UAV) serve as a valuable data source for consistent and continuous monitoring of aboveground biomass. This research focuses on modeling the relationships between AGB and data obtained from various remote sensing sources, including Sentinel-1, Sentinel-2, Landsat 8, and UAV imagery, within pure Scots pine stands in northern T & uuml;rkiye. The study employs multiple linear regression (MLR) and artificial neural networks (ANNs) to establish these relationships. AGB values for each sample plot were calculated using an allometric equation. Backscatter coefficients and band brightness values were extracted from Sentinel-1 imagery, while reflectance values and vegetation indices were generated from Sentinel-2, Landsat 8 OLI, and UAV imagery. Additionally, texture features were computed for varying window sizes (3 x 3, 5 x 5, 7 x 7, 9 x 9, 11 x 11, 13 x 13, and 15 x 15) and orientations (0 degrees, 45 degrees, 90 degrees, and 135 degrees) based on data from Sentinel-2 and Landsat 8 OLI images for each sample plot. The relationships between remote sensing data and AGB were modeled using both MLR and ANN techniques. The findings revealed that the most accurate AGB estimation (R-2=0.82; RMSE = 0.35 ton ha(-)(1)) was achieved using the texture variables derived from the 9 x 9 window size of Sentinel-2 imagery via the ANNs modeling approach, outperforming other image sources and MLR analysis.
dc.description.sponsorshipScientific Research Project Unit of Cankiri Karatekin University [OF211221D08]
dc.description.sponsorshipThis study was funded by the Scientific Research Project Unit of Cankiri Karatekin University (Grant No: OF211221D08).
dc.identifier.doi10.1007/s12145-024-01657-0
dc.identifier.issn1865-0473
dc.identifier.issn1865-0481
dc.identifier.issue1
dc.identifier.scopus2-s2.0-85212445424
dc.identifier.scopusqualityQ2
dc.identifier.urihttps://doi.org/10.1007/s12145-024-01657-0
dc.identifier.urihttps://hdl.handle.net/11486/6796
dc.identifier.volume18
dc.identifier.wosWOS:001380076200006
dc.identifier.wosqualityQ2
dc.indekslendigikaynakWeb of Science
dc.indekslendigikaynakScopus
dc.language.isoen
dc.publisherSpringer Heidelberg
dc.relation.ispartofEarth Science Informatics
dc.relation.publicationcategoryMakale - Uluslararası Hakemli Dergi - Kurum Öğretim Elemanı
dc.rightsinfo:eu-repo/semantics/closedAccess
dc.snmzKA_WOS_20250323
dc.subjectUAV
dc.subjectANNs
dc.subjectNatural forest
dc.subjectForest stand metrics
dc.titleUAV and satellite-based prediction of aboveground biomass in scots pine stands: a comparative analysis of regression and neural network approaches
dc.typeArticle

Dosyalar

Koleksiyon

WoS İndeksli Yayınlar Koleksiyonu
Scopus İndeksli Yayınlar Koleksiyonu