Izvorni znanstveni članak
https://doi.org/https://doi.org/10.5513/JCEA01/25.4.4385
Use of UAVs' multispectral images for sugar beet cultivars discrimination and yield estimation
Vasileios DRIMZAKAS–PAPADOPOULOS
; Department of Surveying Engineering & Geoinformatics, International Hellenic University, I.H.U. Serres Campus, Terma Magnesias str., 62124, Serres, Greece
Konstantinos NTOUROS
; Department of Surveying Engineering & Geoinformatics, International Hellenic University, I.H.U. Serres Campus, Terma Magnesias str., 62124, Serres, Greece
*
Constantine PAPATHEODOROU
; Department of Surveying Engineering & Geoinformatics, International Hellenic University, I.H.U. Serres Campus, Terma Magnesias str., 62124, Serres, Greece
Alexandros KONSTANTINIDIS
; Department of Surveying Engineering & Geoinformatics, International Hellenic University, I.H.U. Serres Campus, Terma Magnesias str., 62124, Serres, Greece
* Dopisni autor.
Sažetak
The significance of crop mapping using remote sensing data is increasingly recognized as a cornerstone for tackling global challenges such as food security and climate change, due to its role in providing accurate and timely information on crop distribution, essential for informed agricultural decision-making. The main objective of this study was to investigate the effectiveness of multispectral UAV imagery for discriminating between sugar beet cultivars and predicting yield. The specific objectives were: i Evaluation of the separability of spectral bands and vegetation indices for 25 sugar beet cultivars using histogram correlation analysis, and ii. Investigating potential correlations between vegetation indices and yield. The results showed the NIR spectral region is prominent followed by Green on both acquisition dates in the S2 control zone in contrast to control zone S1, where Green is the primary spectral region on both acquisition dates. Among vegetation indices, GNDVI demonstrated better separability capability than the other indices (NDVI and RENDVI) in the S2 control zone and on both acquisition dates whereas NDVI performed better results in the S1 control zone and both acquisition dates. Finally, the regression analysis revealed a second-order polynomial equation relating root weight to vegetation pixels (GNDVI) with R2 = 0.34 whereas the average prediction is about 17.62% of the actual value (MAPE). The study shows that the multispectral data have limitations in discriminating between sugar beet cultivars and yield prediction. Further research should be conducted, considering the different phenological stages of the cultivars and multi-annual monitoring.
Ključne riječi
sugar beet; cultivar discrimination; yield estimation; vegetation indices; UAVs multispectral imagery; Python programming language
Hrčak ID:
325004
URI
Datum izdavanja:
23.12.2024.
Posjeta: 0 *