High-resolution Digital Surface Research Articles

High-resolution topography reveals morphological changes of Stromboli volcano following the July 2024 eruption

AbstractThe July 2024 eruption of Stromboli volcano has been characterised by the manifestation, at variable intensity, of the entire repertoire of volcanic events that Stromboli volcano is capable of, and is by far the one that has most changed the morphology of the crater terrace and of the Sciara del Fuoco slope in the last decades. We present the results of an Unoccupied Aircraft System (UAS) survey performed on 14 July 2024 and consisting of 4,988 visible and thermal photographs coupled with Structure-from-Motion photogrammetry that allowed us to produce a high-resolution (0.2 m/pixel) Digital Surface Model (DSM). We documented the profound morphological changes of the Stromboli volcano resulting from the 4–11 July 2024 eruption and obtained elevation and volume change estimates by differencing our survey and a UAS-derived pre-event surface (24 May 2024).

Rockfall mapping and susceptibility evaluation based on UAV high-resolution imagery and support vector machine method

Abstract This study aims to develop a comprehensive method for automated rockfall mapping and susceptibility assessment using unmanned aerial vehicle (UAV) tilt photography and the support vector machine (SVM) algorithm. By employing Jinzhong Town in the mountainous forest landscapes of Southwest China as a case study, we leverage photogrammetry principles and computer vision algorithms to generate high-precision, high-resolution digital surface models (DSMs), and digital orthophoto maps through a UAV remote sensing system. The rockfall inventory is accurately and automatically mapped using the object-based classification method and SVM algorithm. The automated rockfall identification method achieves a 93% accuracy with a Kappa coefficient of 0.7967. Statistical analyses of spatial distribution characteristics reveal a significant correlation between rockfall volume and area with a correlation coefficient (R²) of 0.92316 under logarithmic coordinates. In addition, a power function relationship describes the link between rockfall volume and slope height (R² = 0.87), while the relationship with sliding distance is characterized by a weaker linear correlation (R² = 0.65). Rockfall runout distance also shows a significant linear correlation with slope height (R² = 0.79) but exhibits a less-pronounced association with rockfall volume (R² = 0.58). The SVM model employed to assess rockfall susceptibility indicates high accuracy (area under the curve = 0.896), affirming its efficacy in rockfall susceptibility assessment. Our findings underscore the utility of UAV remote sensing for rockfall information extraction and susceptibility evaluation, particularly in challenging mountainous forest environments characterized by intricate topography and geological complexities.

The effect of terrain on the fine-scale genetic diversity of sub-Antarctic Collembola: A landscape genetics approach.

Biodiversity patterns are shaped by the interplay between geodiversity and organismal characteristics. Superimposing genetic structure onto landscape heterogeneity (i.e., landscape genetics) can help to disentangle their interactions and better understand population dynamics. Previous studies on the sub-Antarctic Prince Edward Islands (located midway between Antarctica and Africa) have highlighted the importance of landscape and climatic barriers in shaping spatial genetic patterns and have drawn attention to the value of these islands as natural laboratories for studying fundamental concepts in biology. Here, we assessed the fine-scale spatial genetic structure of the springtail, Cryptopygus antarcticus travei, which is endemic to Marion Island, in tandem with high-resolution geological data. Using a species-specific suite of microsatellite markers, a fine-scale sampling design incorporating landscape complexity and generalised linear models (GLMs), we examined genetic patterns overlaid onto high-resolution digital surface models and surface geology data across two 1-km sampling transects. The GLMs revealed that genetic patterns across the landscape closely track landscape resistance data in concert with landscape discontinuities and barriers to gene flow identified at a scale of a few metres. These results show that the island's geodiversity plays an important role in shaping biodiversity patterns and intraspecific genetic diversity. This study illustrates that fine-scale genetic patterns in soil arthropods are markedly more structured than anticipated, given that previous studies have reported high levels of genetic diversity and evidence of genetic structing linked to landscape changes for springtail species and considering the homogeneity of the vegetation complexes characteristic of the island at the scale of tens to hundreds of metres. By incorporating fine-scale and high-resolution landscape features into our study, we were able to explain much of the observed spatial genetic patterns. Our study highlights geodiversity as a driver of spatial complexity. More widely, it holds important implications for the conservation and management of the sub-Antarctic islands.

Flood plain inundation modeling with explicit description of land surface macrostructures

Although high-resolution digital surface model (DSM) data derived from lidar surveys can describe land surface macrostructures like trees and buildings, digital terrain model (DTM) data obtained by filtering out these macrostructures are commonly used in flood inundation models. In the present study, it is shown for the first time that DSM data can be used directly in flood inundation models by employing automatically-extracted ridges as breaklines for the generation of geomorphologically-informed meshes (GIMs). Even under the simplifying assumption of impermeable macrostructures, especially when GIM refinement is applied, the use of DSM data in preference to DTM data leads to significant improvement in flood predictions. By comparing simulations and observations for a real flood inundation, it is found that the direct use of 1-m DSM data in place of the related DTM data leads to a 42% improvement in predicted flood area, a 36% improvement in predicted flood areal position, and a 25% improvement in predicted times of travel.

Massive sediment pulses triggered by a multi-stage 130 000 m3 alpine cliff fall (Hochvogel, DE–AT)

Abstract. Massive sediment pulses in catchments are a key alpine multi-risk component. Substantial sediment redistribution in alpine catchments frequently causes flooding, river erosion, and landsliding and affects infrastructure such as dam reservoirs as well as aquatic ecosystems and water quality. While systematic rock slope failure inventories have been collected in several countries, the subsequent cascading sediment redistribution is virtually unaccessed. For the first time, this contribution reports the massive sediment redistribution triggered by the multi-stage failure of more than 130 000 m3 from the Hochvogel dolomite peak during the summer of 2016. We applied change detection techniques to seven 3D-coregistered high-resolution true orthophotos and digital surface models (DSMs) obtained through digital aerial photogrammetry later optimized for precise volume calculation in steep terrain. The analysis of seismic information from surrounding stations revealed the temporal evolution of the cliff fall. We identified the proportional contribution of > 600 rockfall events (> 1 m3) from four rock slope catchments with different slope aspects and their volume estimates. In a sediment cascade approach, we evaluated erosion, transport, and deposition from the rock face to the upper channelized erosive debris flow channel, then to the widened dispersive debris flow channel, and finally to the outlet into the braided sediment-supercharged Jochbach river. We observe the decadal flux of more than 400 000 m3 of sediment, characterized by massive sediment waves that (i) exhibit reaction times of 0–4 years in response to a cliff fall sediment input and relaxation times beyond 10 years. The sediment waves (ii) manifest with faster response times of 0–2 years in the upper catchment and over 2 years in the lower catchments. The entire catchment (iii) undergoes a rapid shift from sedimentary (102–103 mm a−1) to massive erosive regimes (102 mm a−1) within single years, and the massive sediment redistribution (iv) shows limited dependency on rainfall frequency and intensity. This study provides generic information on spatial and temporal patterns of massive sediment pulses in highly sediment-charged alpine catchments.

Coastal Compound Flood Simulation through Coupled Multidimensional Modeling Framework

Coastal watersheds are prone to compound flooding caused by heavy rainfall, storm surge, and high tide. Accurate prediction and modeling of coastal compound floods (CCF) require precise physical representation in numerical models with appropriate numerical approximation associated with multi-driver physical processes for simulating the flood dynamics. To address this, an integrated and tightly coupled modeling approach is implemented to capture the nonlinear and complex processes involved in CCF and interconnectivity among Multidimensional Hydraulics (pipe and channel), Hydrodynamics (2D overland flow), and distributed Hydrologic (MH3) models are developed. The coupled modeling framework focuses on the development of interconnected meshes of the node-link-basin using the Interconnected Channel and Pond Routing (ICPR) model. This framework effectively incorporates the complex drainage network system of tidal creeks, tidal channels, underground sewer networks, and detention ponds of the Charleston Peninsula of South Carolina (SC). The dynamics of the urbanized floodplain in the peninsula are represented using high-resolution LiDAR-derived Digital Elevation Model (DEM) and Digital Surface Model (DSM). The overland flow is simulated using energy balance, momentum balance, and diffusive wave methods. The performance of the CCF model is tested for the 2015 SC historical flood and a high tide flood event. The momentum balance-based CCF model shows 98.35% accuracy in identifying street-level flooding locations. Furthermore, the model shows that incorporating DSM processing in the current study improves the accuracy of urban flood simulation by approximately 15% to 33% compared to LiDAR DEM. It is revealed that the momentum balance method outperforms to simulate the flood dynamics. Consequently, this study contributes to the modeling of CCF by developing an integrated MH3 system that enhances the physical representation of coastal hydrology at a fine scale.

Analyzing Debris Flow: Topographical Data and Discharge Rate Study

This study used an unmanned aerial vehicle (UAV) for aerial photogrammetry at the Seonseri debris flow site in Gokseong-gun, triggered by heavy rainfall in August 2020. Survey data facilitated the generation of orthographic images and high-resolution digital surface mode (DSM). Finite difference method (FDM) numerical analysis simulated debris flow, exploring the impact of discharge rate and topographic data among input variables. Applying the rational formula for determining debris flow discharge rate emphasized the significance of maximum rainfall intensity just before the event, with high estimated values for runoff coefficient and soil concentration. The study highlighted significant variations in debris flow simulation results based on topographical data accuracy, emphasizing the crucial role of precise site-specific topographical data.

The March 2023 UAS-based high-resolution Digital Surface Model and orthomosaic of the NE flank of Stromboli volcano (Sicily, Italy)

Stromboli is a volcanic island in a persistent state of activity, located in the Tyrrhenian Sea off the northern coast of Sicily. During the night of 25 and 26 May 2022, a massive human-caused wildfire destroyed most of the vegetation cover on the NE flank of the island, just above the main village. On 12 August 2022, a particularly heavy rainfall event remobilized the loose volcaniclastic deposits that covered the burned volcanic flank, no longer protected by the vegetation. This event triggered several debris flows that were channeled by the roads and flooded several streets and buildings, causing severe damage to the village. In late-March 2023, just before the large spring vegetation growth, we conducted an Unmanned Aerial System (UAS) photogrammetric campaign over a sector of the NE flank of Stromboli Island, to acquire data on an area massively affected by the wildfire first and by the debris flows later. Here we present and share with the scientific community and civil authorities the results of this UAS campaign, which consists of a 1.4 km2 wide 10 cm-resolution Digital Surface Model (DSM) and 1.6 cm-resolution orthomosaic. These data clearly show the dramatic consequences of the 2022 tragic events at Stromboli. We also produced an elevation difference map by comparing the 2023 DSM here generated and the 2012 LiDAR DEM to provide a first overview of the thickness of the deposits that were removed from the Stromboli NE flank.

Using high-resolution images to analyze the importance of crown size and competition for the growth of tropical trees

The influence of canopy structure on tropical tree growth has been scantly studied because of the difficulties making field measurements in these dense multi-layered ecosystems. The recent advent of unmanned aerial vehicles (UAVs), has made it easier to collect canopy data, so offering a way to gain a better understanding of forest productivity and thereby improve forest management. In this study, we assessed tree growth prediction using UAV-derived crown measurements as an alternative for field data.Four experimental 9 ha plots were sampled in two forest sites, Yoko in the Democratic Republic of the Congo and Loundoungou in the Republic of Congo. Field inventories were made between 2015 and 2020. For each tree, we computed the diameter increment (DBHI) using censuses and diameter-based competition indices (diameter-based CIs) using the first census. High-resolution orthoimages and digital surface models were acquired with UAVs in 2016 and 2018 in the two sites. They gave estimates of crown characteristics (size, relative elevation, shape) and crown-based competition indices (crown-based CIs). Co-recorded UAV and field measurements were obtained for 1558 trees. The diameter increment of these trees was then modelled using supervised component generalized linear regression, and 20 % of trees were kept for cross-validation.Combined field and UAV data predicted tree DBHI twice better than either taken separately. Diameter at breast height (DBH) and crown area (CA) were found to be complementary predictors. Crown-based CIs significantly improved predictions of models already containing DBH and CA. Adding diameter-based CIs to models containing DBH, CA, and crown-based CIs only marginally improved growth predictions, showing that tree competition can be well-described with UAV data. The model calibrated at one site predicted the growth at the other site well, suggesting that a general model could be devised for multiple sites. Growth variance was better explained in the site (Yoko) where the crown density was higher and the crown smaller. Further data are now needed from multiple sites with ranging stand structures and compositions to build a general model.

Benggang segmentation via deep exchanging of digital orthophoto map and digital surface model features

Benggang is a typical fragmented erosional landscape in southern and southeastern China, posing significant risk to the local residents and economic development. Therefore, an efficient and accurate fine-grained segmentation method is crucial for monitoring the Benggang areas. In this paper, we propose a deep learning-based automatic segmentation method for Benggang by integrating high-resolution Digital Orthophoto Map (DOM) and Digital Surface Model (DSM) data. The DSM data is used to extract slope maps, aiming to capture primary morphological features. The proposed method consists of a dual-stream convolutional encoder-decoder network in which multiple cascaded convolutional layers and a skip connection scheme are used to extract morphological and visual features of the Benggang areas. The rich discriminative information in the DOM and slope data is fused by a channel exchanging mechanism that dynamically exchanges the most discriminative features from either the DOM or DSM stream according to their importance at the channel level. Evaluation experiments were conducted on a challenging dataset collected from Guangdong Province, China, and the results show that the proposed channel exchanging network based deep fusion method achieves 84.62% IoU in Benggang segmentation, outperforming several existing unimodal or multimodal baselines. The proposed multimodal segmentation method greatly improves the efficiency of large-scale discovery of Benggang, and thus is important for the management and restoration of Benggang in southern and southeastern China, as well as the monitoring of other similar erosional landscapes.

Urban thermal map design considerations: color, shading, and resolution

ABSTRACT While satellite-based remote sensing techniques are often used for studying and visualizing the heat distribution in cities, they are limited in terms of spatial resolution, view bias, and revisit times. In comparison, modern Unmanned Aerial Vehicles (UAVs) equipped with infrared sensors allow very fine-scale (cm) data to be collected over smaller areas. We present a user study (n = 66) that outlines how satellite and drone-sourced thermal pseudo-color images compare in terms of map reading performance on three representative map reading tasks, how the choice of colormap affects map reading, and how a new shading augmentation to thermal maps based on high-resolution digital surface models can support interaction. Additionally, users provided explicit preferences indicating an inclination toward the shading augmentation, for the recently designed rainbow-style colormap turbo, and the cmocean thermal/FLIR ironbow colormap. However, we detail how user preferences and map reading performance are not always well aligned and are linked to issues of information content and visual clutter.

Building a bimodal landscape: bedrock lithology and bed thickness controls on the morphology of Last Chance Canyon, New Mexico, USA

Abstract. We explore how rock properties and channel morphology vary with rock type in Last Chance Canyon, Guadalupe Mountains, New Mexico, USA. The rocks here are composed of horizontally to near-horizontally interbedded carbonate and sandstone. This study focuses on first- and second-order channel sections, where the streams have a lower channel steepness index (ksn) upstream and transition to higher ksn values downstream. We hypothesize that differences in bed thickness and rock strength influence ksn values, both locally by influencing bulk bedrock strength and also nonlocally through the production of coarse sediment. We collected discontinuity intensity data (the length of bedding planes and fractures per unit area), Schmidt hammer rebound measurements, and measured the largest boulder at every 12.2 m elevation contour to test this hypothesis. Bedrock and boulder mineralogy were determined using a lab-based carbonate dissolution method. High-resolution orthomosaics and digital surface models (DSMs) were generated from drone and ground-based photogrammetry. The orthomosaics were used to map channel sections with exposed bedrock. The United States Geological Survey (USGS) 10 m digital elevation models (DEMs) were used to measure channel slope and hillslope relief. We find that discontinuity intensity is negatively correlated with Schmidt hammer rebound values in sandstone bedrock. Channel steepness tends to be higher where reaches are primarily incising through more thickly bedded carbonate bedrock and lower where more thinly bedded sandstone is exposed. Bedrock properties also influence channel morphology indirectly, through coarse sediment input from adjacent hillslopes. Thickly bedded rock layers on hillslopes erode to contribute larger colluvial sediment to adjacent channels, and these reaches have higher ksn values. Larger and more competent carbonate sediment armors both the carbonate and the more erodible sandstone and reduces steepness contrasts across rock types. We interpret that in the relatively steep, high-level ksn downstream channel sections, the slope is primarily controlled by the coarse alluvial cover. We further posit that the upstream low-level ksn reaches have a base level that is fixed by the steep downstream reaches, resulting in a stable configuration, where channel slopes have adjusted to lithologic differences and/or sediment armor.

Novel UAV Flight Designs for Accuracy Optimization of Structure from Motion Data Products

Leveraging low-cost drone technology, specifically the DJI Mini 2, this study presents an innovative method for creating accurate, high-resolution digital surface models (DSMs) to enhance topographic mapping with off-the-shelf components. Our research, conducted near Jena, Germany, introduces two novel flight designs, the “spiral” and “loop” flight designs, devised to mitigate common challenges in structure from motion workflows, such as systematic doming and bowling effects. The analysis, based on height difference products with a lidar-based reference, and curvature estimates, revealed that “loop” and “spiral” flight patterns were successful in substantially reducing these systematic errors. It was observed that the novel flight designs resulted in DSMs with lower curvature values compared to the simple nadir or oblique flight patterns, indicating a significant reduction in distortions. The results imply that the adoption of novel flight designs can lead to substantial improvements in DSM quality, while facilitating shorter flight times and lower computational needs. This work underscores the potential of consumer-grade unoccupied aerial vehicle hardware for scientific applications, especially in remote sensing tasks.

Mapping intertidal oyster farms using unmanned aerial vehicles (UAV) high-resolution multispectral data

In France, oyster aquaculture has been historically developed in intertidal zones, with shellfish farming areas covering much of the Atlantic coast. Monitoring these off-bottom cultures where oysters are grown in plastic mesh-bags set on trestle tables is mandatory for the maritime administration to check compliance with a Structural Plan Document (SPD), while also being important for stock assessment in relation to carrying capacity issues. However, traditional monitoring methods are time-consuming, labor-intensive, and inefficient in covering large intertidal areas. In this study, we used a new GIS-based analytical method to assess the potential of high-resolution Unmanned Aerial Vehicle (UAV) multispectral data to retrieve spatial information on oyster-farming structures using Bourgneuf Bay (France) as a case-study. A non-parametric machine learning algorithm was applied to four UAV flight orthomosaics collected at different altitudes (12, 30, 50, and 120 m) to identify oyster mesh-bags. These supervised classifications achieved overall accuracies above 95% for all tested altitudes. In addition, an accurate distinction of oyster-bag mesh sizes (4, 9 and 14 mm) was obtained for 12–50 m flights, but there was a lower accuracy at 120 m. Across all flights, the 4 mm mesh size was the least well detected (72.14% Producer Accuracy). This information can be used to identify bags with specific mesh-sizes used for spat or adult grow out. Finally, we accurately measured oyster table heights using a high-resolution Digital Surface Model (DSM) derived from Structure from Motion (SfM) photogrammetry. The 50 m flight was suggested as the best compromise to obtain precise measurements of the oyster table heights while covering larger areas than lower altitude flights. This demonstrates that UAV technology can provide a set of spatial variables relevant for shellfish farmers and coastal managers in an efficient, rapid, and non-destructive way to monitor the extent and characteristics of oyster-farming areas regularly.

Assessing the short-term inter-annual growth of the largest documented gully network in South Africa using UAV and SfM methodology

ABSTRACT This research explores intra-gully sediment volume changes of a massive gully network (42 ha) located in the Eastern Cape, RSA, in response to high rainfall events during a three-year study period (2017–2020). Unmanned aerial vehicles (UAVs) and photogrammetric processing were used to generate high resolution digital surface models (DSMs). Volumetric calculations were run to estimate the extent of gully expansion within the system to determine intra-gully changes and sediment erosion over the study period. The gully was categorised into four zones based on predominant structure and processes. These are the Extension Zone, Expansion Zone, Transport/Deposition Zone and the Side Branches. Rainfall data from a rainfall station (10 km away) was used to investigate the correlation between rainfall depth, rainfall intensity, and flow accumulation over two wetter years and one drier year with gully expansion experienced during these periods. The gully network encountered a total gully sediment volume erosion of 104,489 m3 (6.01%) over the study period. Nineteen high to extreme intensity (>20 mm per day) rainfall events produced the runoff necessary for gully expansion during the 2017 and 2018 hydrological years. Notably, lower gully growth was recorded in the Expansion and Transport/Deposition Zones during the subsequent below average hydrological year in 2019–2020.

High-resolution high-accuracy orthophoto map and digital surface model of Forni Glacier tongue (Central Italian Alps) from UAV photogrammetry

ABSTRACT This work presents the high-resolution high-accuracy orthophoto map and the Digital Surface Model of Forni Glacier (Italian Alps). These represent the status of the glacier tongue in mid-August 2022 when surveys were carried out with a DJI Phantom 4 RTK drone. The processing was carried out in Leica Infinity, and a 3 cm orthomosaic and a 20 cm Digital Surface Model were generated and made available for analysis of the current status of the glacier, which shows signs of downwasting, with the occurrence of collapsing areas and a rapidly changing proglacial landscape. This work can also be used as a reference to investigate the glacier evolution, also in light of climate change. Accuracy requirements of the deliverables were ensured by combining Post Processed Kinematic and Structure from Motion integrated with bundle block adjustment, and using Ground Control Points and Check Points, to guarantee redundancy and evaluate the geolocation accuracy and precision.

Co-occurrence of pluvial and fluvial floods exacerbates inundation and economic losses: evidence from a scenario-based analysis in Longyan, China

Hydrodynamics and socio-economic impacts of compound floods as the result of co-occurring pluvial and fluvial floods, have not been well studied, which challenges conventional urban flood risk management that treats different types of floods separately. This study generates a high-resolution land surface digital surface model based on images from unmanned aerial vehicles, and incorporates it into a combined flooding model that integrates the 1D river hydrodynamic model, 1D urban drainage hydrodynamic model and 2D overland flow model, in order to simulate the inundation characteristics of extreme floods when local pluvial floods coincide with upstream fluvial floods, and in a further step to quantify the direct economic losses on urban physical systems. As demonstrated by an empirical study of Longyan, China, the combination of pluvial and fluvial floods leads to substantial exacerbation in inundation extent and depth. Consequently, the joining of pluvial and fluvial floods greatly amplifies economic losses to urban physical systems, indicating that ignoring co-occurrence of different types of floods will potentially lead to underestimation of flood risk and insufficient adaptation. Our study emphasizes the need to account for the co-occurrence of multiple types of floods in flood risk assessment and management, to avoid the shortcomings of fragmented responses.

Measuring landslide-driven ground displacements with high-resolution surface models and optical flow

This study proposes and evaluates the performance of two image matching algorithms applied to hillshades derived from consecutive high-resolution digital surface models (DSMs) to measure surface displacements on landslides. The method is applied on Te Horo, a slow-moving landslide in the Dart Valley of New Zealand’s Southern Alps/Kā Tiritiri o te Moana using pairs of hillshades derived from Airborne and Satellite Photogrammetric Mapping (APM/SPM) of imagery captured in 2018 and 2020. This novel approach uses the consistency of displacement predictions generated from multiple hillshade pairs to gauge match quality and mask unreliable displacement predictions. The study compares a widely used normalised cross-correlation algorithm (NCC) alongside an optical flow approach for image matching. The performances of both algorithms are assessed against manually derived displacements of prominent surface features. We demonstrate the effectiveness of the masking approach as well as the good performance of the optical flow algorithm in delivering dense, accurate displacement measurements efficiently, particularly when high resolution DSMs are available. The results show that the main translational body of the landslide was displaced at rates averaging 25–55 mm day−1 over the 2018–2020 period.

UAV Photogrammetry for Soil Surface Deformation Detection in a Timber Harvesting Area, South Korea

During forest operations, canopy removal results in the soil surface being vulnerable to deformation, negatively impacting soil fertility and water quality. This study utilized unmanned aerial vehicle (UAV) photogrammetry to accurately detect soil surface deformation (SSD). Two-dimensional images were safely collected on a steep slope without real-time kinematics by conducting vertically parallel flights (VPFs). A high-resolution digital surface model (DSM) with a <3 cm resolution was acquired for precise SSD detection. Using DSM of difference (DoD), SSDs were calculated from DSMs acquired in June, July, September, and October 2022. By checking spatial distances at ground control points, errors of DSM alignments were confirmed as only 3 cm, 11.1 cm, and 4 cm from July to June, September to June, and October to June, respectively. From the first month of monitoring, erosion and deposition of approximately 7 cm and 9 cm, respectively, were detected at validation points (VPs). However, from total monitoring, cumulative SSD was assessed as having deposition tendencies at all VPs, even compared to ground truths. Although UAV photogrammetry can detect SSDs, spatial distortion may occur during UAV surveys. For vegetation growth issues, UAV photogrammetry may be unable to capture data on the soil surface itself.

Mapping Tree Species Using CNN from Bi-Seasonal High-Resolution Drone Optic and LiDAR Data

As the importance of forests has increased, continuously monitoring and managing information on forest ecology has become essential. The composition and distribution of tree species in forests are essential indicators of forest ecosystems. Several studies have been conducted to classify tree species using remote sensing data and machine learning algorithms because of the constraints of the traditional approach for classifying tree species in forests. In the machine learning approach, classification accuracy varies based on the characteristics and quantity of the study area data used. Thus, applying various classification models to achieve the most accurate classification results is necessary. In the literature, patch-based deep learning (DL) algorithms that use feature maps have shown superior classification results than point-based techniques. DL techniques substantially affect the performance of input data but gathering highly explanatory data is difficult in the study area. In this study, we analyzed (1) the accuracy of tree classification by convolutional neural networks (CNNs)-based DL models with various structures of CNN feature extraction areas using a high-resolution LiDAR-derived digital surface model (DSM) acquired from a drone platform and (2) the impact of tree classification by creating input data via various geometric augmentation methods. For performance comparison, the drone optic and LiDAR data were separated into two groups according to the application of data augmentation, and the classification performance was compared using three CNN-based models for each group. The results demonstrated that Groups 1 and CNN-1, CNN-2, and CNN-3 were 0.74, 0.79, and 0.82 and 0.79, 0.80, and 0.84, respectively, and the best mode was CNN-3 in Group 2. The results imply that (1) when classifying tree species in the forest using high-resolution bi-seasonal drone optical images and LiDAR data, a model in which the number of filters of various sizes and filters gradually decreased demonstrated a superior classification performance of 0.95 for a single tree and 0.75 for two or more mixed species; (2) classification performance is enhanced during model learning by augmenting training data, especially for two or more mixed tree species.