Spatially robust pixel‐based mapping of biotic stress in Pinus nigra using unmanned aerial vehicle multispectral imagery and Random Forest

Abstract BACKGROUND Biotic stress threatens forest health, creating a need for rapid monitoring methods that complement field surveys. This study compared four original multispectral bands, five vegetation indices, and their combination for segmentation‐free, pixel‐based Random Forest mapping of healthy and unhealthy Pinus nigra canopies in Bingöl, Türkiye. Field and laboratory observations confirmed Dothistroma spp . and Tomicus spp . within the stand. RESULTS Unmanned aerial vehicle (UAV) imagery with a ground sampling distance of 2.31 cm was evaluated using identical five‐block spatial cross‐validation partitions comprising 113 georeferenced trees (55 healthy and 58 unhealthy), each represented by mean predictor values from a manually delineated within‐crown sampling polygon. The combined model achieved the highest pooled overall accuracy (98.23%; Kappa = 0.9646). The vegetation‐index model achieved 97.35% accuracy and the highest area under the receiver operating characteristic curve (ROC‐AUC, 0.9991), with Optimized Soil‐Adjusted Vegetation Index (OSAVI) showing the greatest mean permutation importance. Exact McNemar tests detected no significant differences among models (all P = 1.000). The vegetation‐index model was retained as a parsimonious alternative to the combined model and fitted using 200 reference polygons to generate the operational map. Cross‐validated performance does not constitute independent validation of this final map. CONCLUSION Comparing predictor representations under a common spatial validation framework supports transparent assessment of UAV‐based canopy‐health mapping without crown segmentation. The approach can guide local field inspections but does not identify individual causal agents. Findings represent within‐flight discrimination in one stand; absent reflectance‐panel calibration, broader application requires standardized radiometric calibration and independent spatial and temporal validation. © 2026 Society of Chemical Industry.

Authors

Institutions

Publication Details

Journal
Pest Management Science
Published
2026-11-15
DOI
https://doi.org/10.1002/ps.71498
Primary Topic
Remote Sensing in Agriculture
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Spatially robust pixel‐based mapping of biotic stress in Pinus nigra using unmanned aerial vehicle multispectral imagery and Random Forest

Alaaddin Yüksel, Alperen Meral, Osman Eren Arslan, Işıl Saraç Sivrikaya
Pest Management Science
Remote Sensing in Agriculture
article

Spatially robust pixel‐based mapping of biotic stress in Pinus nigra using unmanned aerial vehicle multispectral imagery and Random Forest

Alaaddin Yüksel, Alperen Meral, Osman Eren Arslan, Işıl Saraç Sivrikaya
article en

Abstract

Abstract BACKGROUND Biotic stress threatens forest health, creating a need for rapid monitoring methods that complement field surveys. This study compared four original multispectral bands, five vegetation indices, and their combination for segmentation‐free, pixel‐based Random Forest mapping of healthy and unhealthy Pinus nigra canopies in Bingöl, Türkiye. Field and laboratory observations confirmed Dothistroma spp . and Tomicus spp . within the stand. RESULTS Unmanned aerial vehicle (UAV) imagery with a ground sampling distance of 2.31 cm was evaluated using identical five‐block spatial cross‐validation partitions comprising 113 georeferenced trees (55 healthy and 58 unhealthy), each represented by mean predictor values from a manually delineated within‐crown sampling polygon. The combined model achieved the highest pooled overall accuracy (98.23%; Kappa = 0.9646). The vegetation‐index model achieved 97.35% accuracy and the highest area under the receiver operating characteristic curve (ROC‐AUC, 0.9991), with Optimized Soil‐Adjusted Vegetation Index (OSAVI) showing the greatest mean permutation importance. Exact McNemar tests detected no significant differences among models (all P = 1.000). The vegetation‐index model was retained as a parsimonious alternative to the combined model and fitted using 200 reference polygons to generate the operational map. Cross‐validated performance does not constitute independent validation of this final map. CONCLUSION Comparing predictor representations under a common spatial validation framework supports transparent assessment of UAV‐based canopy‐health mapping without crown segmentation. The approach can guide local field inspections but does not identify individual causal agents. Findings represent within‐flight discrimination in one stand; absent reflectance‐panel calibration, broader application requires standardized radiometric calibration and independent spatial and temporal validation. © 2026 Society of Chemical Industry.

Pest Management Science
Fırat University (TR), Bingöl University (TR)
Remote Sensing in Agriculture
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon