Assessing a novel modelling approach with high resolution UAV imagery for monitoring health status in priority riparian forests

Abstract

Background Black alder (Alnus glutinosa) forests are in severe decline across their area of distribution due to a disease caused by the soil-borne pathogenic Phytophthora alni species complex (class Oomycetes), “alder Phytopththora”. Mapping of the different types of damages caused by the disease is challenging in high density ecosystems in which spectral variability is high due to canopy heterogeneity. Data obtained by unmanned aerial vehicles (UAVs) may be particularly useful for such tasks due to the high resolution, flexibility of acquisition and cost efficiency of this type of data. In this study, A. glutinosa decline was assessed by considering four categories of tree health status in the field: asymptomatic, dead and defoliation above and below a 50% threshold. A combination of multispectral Parrot Sequoia and UAV unmanned aerial vehicles -red green blue (RGB) data were analysed using classical random forest (RF) and a simple and robust three-step logistic modelling approaches to identify the most important forest health indicators while adhering to the principle of parsimony. A total of 34 remote sensing variables were considered, including a set of vegetation indices, texture features from the normalized difference vegetation index (NDVI) and a digital surface model (DSM), topographic and digital aerial photogrammetry-derived structural data from the DSM at crown level. Results The four categories identified by the RF yielded an overall accuracy of 67%, while aggregation of the legend to three classes (asymptomatic, defoliated, dead) and to two classes (alive, dead) improved the overall accuracy to 72% and 91% respectively. On the other hand, the confusion matrix, computed from the three logistic models by using the leave-out cross-validation method yielded overall accuracies of 75%, 80% and 94% for four-, three- and two-level classifications, respectively. Discussion The study findings provide forest managers with an alternative robust classification method for the rapid, effective assessment of areas affected and non-affected by the disease, thus enabling them to identify hotspots for conservation and plan control and restoration measures aimed at preserving black alder forests.

Investigating the Correlation between Multisource Remote Sensing Data for Predicting Potential Spread of Ips typographus L. Spots in Healthy Trees

“…This was likely due to the similar chlorophyll-a,b absorption and reflectance properties of the cellular structure of the leaves, especially in the NIR band. Our results were consistent with the results of several other studies that emphasized the importance of VIs in predicting health status classes [27,45,46,[50][51][52][53].…”

UAV-Based Forest Health Monitoring: A Systematic Review

“…Other investigated pathogen effects were caused by bacteria (Ralstonia sp.) [159] and oomycetes represented by Phytophthora cinnamomi [109,115] and Phytophthora alni [89,119]. Acute oak decline was also reported [123,158], which is likely to be caused by an interactive bacterial pathobiome [165].…”

“…Other architectures related to ANNs were multilayer perceptrons [99] and autoencoders [144]. A comparison between different machine learning classifiers was also carried out [119,130,138,142,144,156,157]. The most comprehensive comparative analysis was performed by Wu et al [111].…”

“…The orthomosaic is by far the most widely used product when deploying imaging sensors, particularly in spectral analysis for assessing the health of trees. In some cases, spectral data were enriched with structural information derived from point clouds, DSMs or CHMs based on either LiDAR [116][117][118] or image data [96,102,119,120], further improving classification results. In a few papers, the raw drone images were directly analyzed without further processing [44,108,[121][122][123][124][125][126].…”

The Use of UAV-Acquired Multiband Images for Detecting Rockfall-Induced Injuries at Tree Crown Level

“…Thus far, spectral data for the analysis of tree vitality in forestry were mostly focused on assessing the vitality of the forest in relation to the dynamics of biotic natural disturbances, such as diseases, insect outbreaks (especially damage and early detection of bark beetles) [31,33,[42][43][44][45][46][47][48], and meteorological disasters, such as damage by hurricanes [49]. In this paper, we wanted to investigate the use of the UAV-acquired multiband images for studying the early detection of trees or parts of forest stands where tree vitality is affected due to geomorphologic abiotic disturbances (e.g., rockfalls, avalanches, and debris slides).…”