Mountain forests provide essential ecosystem services, including protection against snow avalanches. Severe disturbances such as windthrow and bark beetle outbreaks generate large amounts of lying deadwood, altering forest structure and affecting avalanche protection. Climate change and legacy effects of past land use are increasing the frequency and severity of such disturbances, while management resources remain limited, highlighting the need for objective decision support tools (DSTs) for evidence-based post-disturbance forest management. We present a spatially explicit DST for deriving spatial indicators related to the protective effect of lying deadwood against avalanche release by integrating terrain, forest structure, and deadwood characteristics. Using high-resolution UAV-derived data, the DSTderivesdeadwoodstructure,models winter terrain under increasing snow depths, and identifies critical snow depths at which surface roughness is reduced. Comparisons with snow-on orthophotos confirm the robustness of the modeled winter terrain. Application across different UAV sensor systems, from low-cost photogrammetry to LiDAR, demonstrates the framework’s broad applicability, showing that even low-cost systems provide suffi ciently accurate results. Analyses across multiple Alpine study sites and post-disturbance management strategies reveal that both the amount and spatial configuration of deadwood control the protective effect. Unmanaged areas maintain the highest protection, requiring snow depths exceeding ∼1.6 m for potential avalanche release, with only minor reductions over a six-year period (∼15% in total). Complete deadwood removal considerably reduces protection, whereas partial removal still provides substantial protection under low-snow conditions. Our open-source DST provides a transparent and practitioner-oriented indicator framework for assessing the protective effects of lying deadwood against snow avalanches under changing disturbance regimes.

For the development of accurate shallow landslide (translational debris and earth slides with a depth < 2 m) susceptibility assessments and further hazard or risk analyses, it is essential that complete and accurate landslide inventory data is available. Various methods are applied for the construction of shallow landslide inventories. However, it is known that the most used methods underreport landslides in forests, e.g. with visual interpretation of satellite/aerial imagery and manual mapping of landslides during field visits. To address this issue, several studies have instead used topographic Light Detection and Ranging (LiDAR) data to create their landslide inventories. These studies showed that landslides under forest cover can be mapped using topographic LiDAR, as LiDAR can penetrate the vegetation cover. The methods used in these studies can be divided into (1) methods using raster data derived from filtered LiDAR point-cloud data and (2) methods working directly on point-cloud datasets. The benefit of the raster-based methods is their computational speed and scalability, while point-cloud based methods are difficult to apply to larger areas, due to their high computational requirements, but have a greater measurement accuracy (e.g., landslide depth). This difference in accuracy is especially important for the mapping of shallow landslides, which often leave only limited traces in the landscape. This study investigates how both methods can be combined to derive a semi-automatic workflow for mapping shallow landslides using LiDAR data that is accurate and scalable. The investigation focusses on mapping shallow landslides under forest, and on how the derived workflow for mapping landslides needs to be adapted to forested and non-forested areas. In a first step, potential landslide-prone areas are identified using the difference of pre- and post-event digital terrain models, an after-event digital terrain model and their related topographic derivatives such as the roughness coefficient and slope. In the next step, the identified areas are segmented and man-made topographic changes are removed, before they are further analyzed with a more accurate mapping technique using point-cloud data from the multiscale model-to-model cloud comparison (M3C2) algorithm. In addition to the M3C2 distances, the point-cloud based mapping will also make use of 3D shape features describing point location and orientation to increase the accuracy and robustness of the topographic change detection and estimation. The scalability of the workflow is tested by applying the workflow to several areas in the Tyrolean Alps (Austria). First results, derived with a logistic regression model using the raster-based derivatives, show a distinct difference in the feature importance of the topographic derivatives when forested and non-forested areas are compared. In addition, the performance of the model also greatly benefits from a separate training in forested and non-forested areas, with an increase in the Area Under the Curve (AUC) value from 0.84 to 0.89 for, respectively, unseparated and separated training.

Mountain forests provide various ecosystem services, including the protection against snow avalanches, which is essential for Alpine communities. However, storms, one of the primary drivers of large-scale forest disturbances in protective forests, can change the forest structure and thus the protective effect against avalanche formation and release. This can potentially lead to a protection gap, where the forest cannot fulfil its protective function. The assessment of the remaining protective effect of these areas against snow avalanches is crucial for decisions regarding the most effective post-disturbance forest and risk management. Previous studies have shown that unmanaged windthrow areas often exhibit a high protective effect against avalanches as the lying stems and root plates lead to a high roughness and prevent the formation of spatially continuous weak layers, but their protective effect changes over time due to decomposition. However, an objective, reliable, and easy-to-apply assessment for monitoring the protective effect is still lacking. Building on a recent study that introduced protective effect indices in windthrow areas derived from drone-based photogrammetric data, we refined and integrated this approach into a semi-automated and comprehensive framework for supporting the assessment of said protective effect of windthrow areas against snow avalanche release. The framework includes 1) the processing of a dense point cloud representing the deadwood structure, 2) the detection of remaining standing trees and their crowns, 3) the determination of the critical snow depth required to cover the deadwood and reduce surface roughness below a defined threshold favourable for avalanche release, 4) adding a relevant snow slab thickness, and 5) assessing the return period of such snow depths. The output of this framework are spatial protective effect indices identifying critical zones for potential avalanche release. We demonstrate the applicability of the framework in various windthrow case study sites, exhibiting different characteristics in severity, deadwood structure, number of standing trees and slope steepness. The results show that the return period for the snow depth required for potential avalanche release is mostly above 30 years but locally varies strongly depending on the existing deadwood structure. This study is a crucial step toward providing an objective decision-support tool for practitioners and decision-makers to effectively manage windthrow areas in protective forests against snow avalanches.

Global environmental changes are widely visible in mountainous areas. These changes can influence the susceptibility of critical infrastructure to well documented and novel natural hazards, cascading as well as compounding risks and potentially threaten cross-border mobility. Within the scope of the EU Horizon project PARATUS, the risk in multiple application case study sites is evaluated regarding past events and future scenarios. Here, we focus on the Brenner Corridor (Austria–Italy), one of the main transit routes through the Alps for both road and rail traffic. Compounding events could potentially lead to a complete blockage of the Brenner Corridor. This demonstrates the importance of a better understanding of regional conditions to develop adjusted adaptation strategies with regard to extreme events, such as extreme wind, floods, rockfall, forest fires, landslides, or snow avalanches. In order to investigate the variety of possible challenges, participatory workshops, scenario modelling and impact chains are developed. Both historical disaster events and future scenarios are analysed. Preliminary results show the great variety of hazards and potential impacts resulting from changing environmental conditions, and the complex interaction of multiple hazards in the Brenner Corridor. Generally, stakeholders are very conscious of these challenges, have well-established concepts in place to face single hazards. However, especially the handling of compounding events and cross-border interaction still has room for improvement. The implementation of impact chains and scenario modelling can help practitioners to find adaptation and mitigation strategies for compounding and cascading hazards. Further work includes scenario modelling on different spatial and temporal scales.

Understanding the dynamics of snow avalanches is crucial for predicting their destructive potential and mobility. To gain insight into avalanche dynamics at a particle level, the AvaNode in-flow sensor system was developed. These synthetic particles, equipped with advanced and affordable sensors such as an inertial measurement unit (IMU) and global navigation satellite system (GNSS), travel with the avalanche flow. This study focuses on assessing the feasibility of the in-flow measurement systems. The experiments were conducted during the winter seasons of 2021–2023, both in static snow cover and dynamic avalanche conditions of medium-sized events. Radar measurements were used in conjunction with the particle trajectories and velocities to understand the behaviour of the entire avalanche flow. The dynamic avalanche experiments allowed to identify three distinct particle flow states: (I) initial rapid acceleration, (II) a steady state flow with the highest velocities (9–17 ms−1), and (III) a longer deceleration state accompanied by the largest measured rotation rates. The particles tend to travel towards the tail of the avalanche and reach lower velocities compared to the frontal approach velocities deduced from radar measurements (ranging between 23–28 ms−1). The presented data give a first insight in avalanche particle measurements.

A detailed knowledge of avalanche dynamics is crucial to optimize flow models that allow avalanche simulation tools to be effectively used for dimensioning mitigation measures or identifying endangered terrain. There are different ways to observe the dynamics in an avalanche during the flow. It can be achieved with remote sensing approaches or fixed sensor systems that interact with the flow. In this Abstract we introduce an inflow sensor system, the so called AvaNodes that are equipped with a variety of sensors, investigating the potential of Global Navigation Satellite System (GNSS) modules. The AvaNode is a cube with 16 cm side length. It is designed to flow in the avalanche and obtain GNSS position and velocity, inertial measurement unit (IMU) based accelerations, angular velocities and the magnetic flux densities, and temperature by means of an infrared thermometer. The utilized GNSS modules are from the ublox CAM-M8 series, that have a position accuracy of 2 m and velocity accuracy of 0.05 m/s, according to the datasheet. To estimate the position accuracy of the AvaNode while covered with snow, experiments were performed with the AvaNode buried in snow at different depths at a known location. Results show that the position accuracy is highly dependent on the number of satellites that the module currently tracks, ranging between 2 and 10 meters. To estimate the GNSS velocity accuracy while the AvaNode is covered with snow, a dynamic experiment with moving sensors was performed. The AvaNode was transported on a sledge while it was buried in 10 and 20 cm of snow. An accuracy in the range of 0.5 m/s was observed, allowing to potentially investigate the dynamics in real avalanches. The influence of burial or snow cover depth did not show conclusive influence on the results and requires further investigation. In 2021 this inflow sensor system was used in two avalanche experiments, on March 15 and 16, obtaining start and end positions, as well as promising GNSS velocities. On March 15 one AvaNode was transported by an avalanche, where the GNSS velocity shows a maximum of 15 m/s and a duration of 50 seconds of the avalanche. On March 16 two AvaNodes were picked up by an avalanche, both showing similar velocity distributions, with a maximum velocity of 17 and 13 m/s.

Abstract. The European Space Agency SnowSAR instrument is a side looking, dual polarized (VV/VH), X/Ku band synthetic aperture radar (SAR), operable from a small aircraft. Between 2010 and 2013, the instrument was deployed at several sites in Northern Finland, Austrian Alps, and northern Canada. The purpose of the airborne campaigns was to measure the backscattering properties of snow-covered terrain to support the development of snow water equivalent retrieval techniques using SAR. SnowSAR was deployed in Sodankylä, Northern Finland for a single flight mission in March 2011 and twelve missions at two sites (tundra and boreal forest) in the winter of 2011–2012. Over the Austrian Alps, three flight missions were performed between November 2012 and February 2013 over three sites located in different elevation zones, representing a montane valley, Alpine tundra, and a glacier environment. In Canada, a total of two missions were flown in March and April 2013, over sites in the Trail Valley Creek watershed, Northwest Territories, representative of the tundra snow regime. This paper introduces the airborne SAR data, as well as coincident in situ information on land cover, vegetation and snow properties. To facilitate easy access to the data record the datasets described here are deposited in a permanent data repository (https://doi.pangaea.de/10.1594/PANGAEA.933255; Lemmetyinen et al., 2021). A temporary link to access the data without login information is provided for reviewers of this manuscript: https://www.pangaea.de/tok/e8c562c3c8a15ac34daa83d00c76fcb347330884.

We present object protection forest maps for rockfall, shallow landslides and snow avalanches, which were generated within the Interreg Alpine Space project GreenRisk4ALPs with the runout model Flow-py. Six Alpine regions with varying sizes from 45 km2 to 2250 km2, and topographies, from steep valleys of Val Ferret in Italy to the German Alpine foothills in Oberammergau, were modeled.The term direct object protection forest is used for forests that protect objects in developed areas against gravitational natural hazards. That is, a direct object protection forest can only be assigned, if an object is endangered and a direct link between the precise locations of the hazard process area and the object can be established. The two main protective effects forests can have against gravitational natural hazards are 1) to reduce the release probability, or 2) to reduce the magnitude of an event, the effectiveness of both is dependent on forest structure. In addition, the degree to which the forest reduces the energy (magnitude) of the hazard also depends on the speed of the mass. If the magnitude/speed of a hazard process is too high, the forest will be destroyed. The location of a forest therefore determines its protective effect in two ways. First, high elevations and steep terrain (over 45°) will produce a weaker structure and be less effective against gravitational natural hazards compared to lower elevation gentle sloped terrain. Second, the energy of the hazard will be lower closer to the hazard’s release and runout areas than in the middle of the process path.Based on these relationships, we generated two types of object protection forest maps:(i) maps that highlight existing direct object protection forest(ii) maps that show where direct object protection forests have or would have the highest potential to either reduce release probability or reduce the energy of the hazardThe Flow-py model was used to model the routing and stopping of the three hazards and to establish the link between endangered objects and the hazard process areas. Input data are digital elevation models (10 m resolution) and locations of release areas as well as a GIS layer containing locations and types of objects, which is required for a custom plugin. The Back-calculation plugin was used with the Flow-py model to identify areas on the terrain (release areas, transit paths and deposition areas) that are associated with endangered infrastructure. To obtain the first maps, the model outputs were overlaid with digital maps of existing forest areas to identify direct object protection forest. The second map was produced by using the same model outputs and digital terrain models to identify areas in the process paths where the modelled hazard energy was low and effective protection forest can grow.The presented maps can help to support decisions and prioritize interventions in risk-based protection forest and ecosystem-based integral natural hazard risk management in the Alpine Space.

Snow avalanches are natural disturbances that can cause substantial damage to forests, and endanger people and material assets. Knowledge of past avalanches is crucial for forest management and planning technical mitigation measures. Dendrogeomorphology can provide information on previous disturbances, for example tree damages, caused by avalanches in forested terrain. By analysing the past growth of trees, both temporal and spatial reconstructions of the avalanche activity in forests are possible. We use a dendrogeomorphological approach to study the past avalanche activity on an avalanche path above the city of Innsbruck in Austria. The area is of high importance for recreation (e.g. hiking, biking and skiing) as well as avalanche mitigation. Protection forest and technical protection measures are already in place (breaking mounds, catching and deflection dams) and frequently interact with avalanches. In January 2019, an avalanche with a destructive size of 3 - 4 released above the Arzler Alm mountain hut and caused considerable damage to approx. 25 ha of forest. This event provided us with the opportunity to conduct the present study. We sampled 104 trees along three longitudinal transects at elevation bands of 1200, 1100 and 1000 m a.s.l. covering the damaged area. We furthermore applied a selective sampling scheme below the forest damage along a gully where avalanches that reached the city of Innsbruck had previously been observed. Using an increment borer at least two cores per tree were taken from damaged and undisturbed trees. A mixture of conifers and broadleaved trees (mostly Picea abies (L.) Karst, Fagus sylvatica L. and Abies alba Mill.), as well as old and young trees was selected. In addition, we recorded the exact position of each tree and measured several tree parameters (e.g. diameter at breast height, tree height, damage description). Each core was then prepared following a standard dendrochronological procedure. Tree-rings were counted and ring-width was measured using a stereo microscope and a time-series analysis program (TSAP Win). Additionally, a visual detection of growth reactions (traumatic resin ducts, reaction wood, scars, callus tissue, growth suppression or releases) was performed, and tree-ring series were cross-dated and compared with local reference chronologies. Years with tree-rings showing growth anomalies potentially caused by ecological or climatic factors were discarded as possible avalanche years. Going forward we will compare years with major avalanche events identified by the dendrogeomorphological analysis, with existing extensive archival data and orthophotos. We expect this to confirm known events, but also to provide new information on unknown events. Based on the location of sampled trees, we will furthermore reconstruct the spatial extent of past events to estimate magnitude and frequency of avalanche activity in the area. Our results will also contribute to better predicting size and periodicity of future avalanche events and revealing potential changes in the avalanche regime. This in turn is relevant to calibrating and validating avalanche simulation models as well as for the design of technical and silvicultural protection and mitigation measures, which is especially important for an Alpine city like Innsbruck.

This paper represents the result of the IAEG C35 Commission “Monitoring methods and approaches in engineering geology applications” workgroup aimed to describe a general overview of unmanned aerial vehicles (UAVs) and their potentiality in several engineering geology applications. The use of UAV has progressively increased in the last decade and nowadays started to be considered a standard research instrument for the acquisition of images and other information on demand over an area of interest. UAV represents a cheap and fast solution for the on-demand acquisition of detailed images of an area of interest and the creation of detailed 3D models and orthophoto. The use of these systems required a good background of data processing and a good drone pilot ability for the management of the flight mission in particular in a complex environment.

The use of unmanned aerial vehicles (UAV) for ground surface measurements in natural hazard studies has strongly increased in recent years. Multi-temporal 3D point clouds derived from light detection and ranging (LiDAR) sensors and photogrammetric techniques including structure-from-motion (SfM) and dense image matching (DIM) have become important tools for monitoring the activity of geomorphic processes. However, due to georeferencing errors and measurement inaccuracies, change detection with centimeter precision remains challenging, especially in study areas covered by vegetation. This study aims at quantifying the influence of low vegetation on the vertical uncertainties of 3D point clouds in a study area mostly covered by meadows and pastures with different grass heights. 3D point clouds derived from UAV-SfM and UAV-LiDAR are compared to terrestrial ground surface measurements of a differential global navigation satellite system (dGNSS) receiver in order to quantify the vertical uncertainties and to detect advantages/disadvantages of the different sensors. The results indicate that neither method is able to detect the ground surface under dense low vegetation with centimeter precision, and that surface displacement rates derived from multi temporal analyses can be highly influenced by changes in vegetation height between surveys.

Mountainous areas bring unique challenges for surveying and natural hazard monitoring – inaccessibility, dangerous terrain, snow coverage and line-of-sight problems often make it next to impossible to perform ground-based monitoring or even to provide a good vantage point for close-range sensing (e.g. terrestrial laser scanning (TLS) or terrestrial photogrammetry). Airborne or satellite-based methods are often the only way to gain information about geodynamically active sites. Here, structure-from-motion (SfM) photogrammetry from unmanned aerial vehicle (UAV) imagery in particular can provide an inexpensive and easily implemented monitoring option. The Vigilans research project attempts to evaluate the feasibility of UAV-photogrammetry against more established surveying methods (e.g. in situ data from extensometers or total stations). Our study site Marzellkamm is located in the Central Ötztal Alps of Western Austria. The active rock slope deformation we are monitoring in Vigilans lies at 2450-2850 m asl. on a SE-facing slope. Annual displacement rates of up to 1.5 m/year in the early 2010’s triggered monitoring and research interest. Due to the remote location, mitigation methods were not implemented, but a hiking trails was relocated. Orthoimage photogrammetry and ground-based monitoring instrumentation (extensometers, terrestrial laser scanning, total station measurements combined with GNSS and geodetic surveys) collected data 1971-2019. In the last years, movement along the slope has slowed down considerably. The rather slow current movements provide a valuable challenge for detection, with rates of <0.05 m/year occurring in the more stable upper sections, while the NW section in particular still shows pronounced movement of up to 0.3 m/year. For this reason, Marzellkamm provides excellent evaluation for new methods such as UAV-SfM. In three separate missions between summer 2018 to fall of 2019, UAV-SfM 3D-models of the site were created for displacement rate evaluations; it is planned to continue this monitoring for a total of three years as part of the Vigilans project. Photogrammetric missions were performed in conjunction with total station measurements of more than 30 ground control points. The required level of precision is becoming achievable and affordable with new RTK/PPK-equipped (Real-Time-Kinematics/Post-Processed Kinematics) UAVs. However, evaluating the resulting 3D-- model in terms of movement rates remains non-trivial. The most common algorithm for change detection in point clouds, M3C2, is not well-suited to detect a laterally moving surface as a whole, as it detects changes along the normal orientation of a surface (such as subsidence). Therefore, the point cloud needs to be very selectively reduced, requiring complex filtering operations and expert input as well as expensive software packages. This contribution will present a workflow to simplify such evaluation, based on 2.5D (DEM-based) algorithms such as IMCORR and DoD (Difference-of-DEMs), in comparison with the more complex 3D-pointcloud based processing. The presented workflow is based on Agisoft Metashape and Open-Source software tools QGIS and Saga GIS. It aims to streamline UAV-based surveying work, 3D-model generation and simplified change detection into a repeatable and easily automatable framework. Special emphasis will be put on estimating the quality of the recorded data.

Displacement rates of mountain slope deformations that can affect entire valley mountain flanks are often measured spatially distributed in‐situ without spatial significance. The spatially explicit measurement and recording of time series of slope deformations is a challenge, as the unstable slopes are often disintegrated into several subdomains, which move with different deformation rates. The current state‐of‐the‐art monitoring systems detect slow to very slow deformation rates between mm/a and several m/a. Using the examples of slope deformations in Saalbach‐Hinterglemm and the deep rock slide Marzellkamm in Austria this paper presents the results of terrestrial laser scans, extensometer measurements, Spaceborne InSAR data, unmanned Aerial System Photogrammetry (UAS‐P), and fixed‐point measurements. The different measurements complement each other and are optimally aligned for different application areas. InSAR data can help to identify hot spots on regional and local scale, while UAS‐P enables for spatially high level accuracy in the detection of subdomains moving at different speeds. For local warning systems TLS, extensometers and GBInSAR deliver higher accuracy.

The town of Innsbruck is characterised by the mountain range ‘Nordkette’, which comprises some major avalanche tracks. The Arzler-Alm avalanche, reached the district of Mühlau in 1935. Since then, several generations of defence structures, such as braking mounds, as well as deflection and retention dams have been built. Several well-documented avalanche events allowed observing the effectiveness of the braking mounds in the Arzler-Alm run-out zone. On the 21 January 2018 the Arzler-Alm avalanche released, nearly reaching the retention dam at the bottom of the run-out zone (1000 m a.s.l.). During the descent, the avalanche partly overflowed several braking mounds, located in the avalanche path and the run-out zone. These structures affected the avalanche flow. Aerial photographs were taken from manned and unmanned platforms over the release area and the deposition on 24 January 2018. They were used to map the affected area, thus providing a basis for the evaluation of the effectiveness of these defence structures and representing a well-documented case study for the simulation of dense snow avalanches and their interaction with defence structures. Back calculations of the avalanche event were performed with SamosAT, considering the defence structures in different ways: a) braking mounds represented as part of the natural terrain by including a DTM with high spatial resolution (1 m ground sampling distance); b) areas with higher artificial resistance represent the braking mounds in the simulations. The simulations were compared to a reference simulation, where defence structures were omitted. Both investigated methods showed an influence on flow patterns and led to reduced avalanche velocities in the area next to the Arzler-Alm hut, compared to the reference simulation.

In this study, we investigate the applicability and performance of UAS-based structure-from-motion SfM photogrammetry on very homogenous snow surfaces, under suboptimal illumination conditions, at two alpine test sites in Tschuggen 2000 m a.s.l. close to Davos, Switzerland, and Lizum 2000 m a.s.l. near Innsbruck, Austria. We discuss the topographic and meteorological challenges for flying UAS missions in high-alpine terrain. Additionally, we compare DSMs calculated from the imagery acquired in the visual VIS, λ = 400–700 nm and near-infrared NIR, λ = 700–830 nm parts of the electromagnetic spectrum. We evaluate the resulting DSMs qualitatively and quantitatively by applying: a differential Global Navigation Satellite System GNSS measurements at the Swiss test site, with an expected accuracy better than 0.1 m within x, y, and z directions; b terrestrial laser scanning TLS at the Austrian test site, with an expected accuracy of ± 0.025 m 1σ, along with a distance-dependent error. The results of this study reveal the potential and limitations of UAS-based SfM photogrammetry for applications on snow-covered, alpine terrain in general and in particular the benefit of NIR imagery on the accuracy and precision of the results.

This contribution presents an automated terrestrial laser scanning (ATLS) setup, which was used during the winter 2016/17 to monitor the snow depth distribution on a NW-facing slope at a high-alpine study site. We collected data at high temporal [(sub-)daily] and spatial resolution (decimetre-range) over 0.8 km² with a Riegl LPM-321, set in a weather-proof glass fibre enclosure. Two potential ATLS-applications are investigated here: monitoring medium-sized snow avalanche events, and tracking snow depth change caused by snow drift. The results show the ATLS data’s high explanatory power and versatility for different snow research questions.

Reliable and timely information on the spatio-temporal distribution of snow in alpine terrain plays an important role for a wide range of applications. Unmanned aerial system (UAS) photogrammetry is increasingly applied to cost-efficiently map the snow depth at very high resolution with flexible applicability. However, crucial questions regarding quality and repeatability of this technique are still under discussion. Here we present a multitemporal accuracy and precision assessment of UAS photogrammetry for snow depth mapping on the slope-scale. We mapped a 0.12 km 2 large snow-covered study site, located in a high-alpine valley in Western Austria. 12 UAS flights were performed to acquire imagery at 0.05 m ground sampling distance in visible (VIS) and near-infrared (NIR) wavelengths with a modified commercial, off-the-shelf sensor mounted on a custom-built fixed-wing UAS. The imagery was processed with structure-from-motion photogrammetry software to generate orthophotos, digital surface models (DSMs) and snow depth maps (SDMs). Accuracy of DSMs and SDMs were assessed with terrestrial laser scanning and manual snow depth probing, respectively. The results show that under good illumination conditions (study site in full sunlight), the DSMs and SDMs were acquired with an accuracy of ≤ 0.25 and ≤ 0.29 m (both at 1 σ ), respectively. In case of poorly illuminated snow surfaces (study site shadowed), the NIR imagery provided higher accuracy (0.19 m; 0.23 m) than VIS imagery (0.49 m; 0.37 m). The precision of the UAS SDMs was 0.04 m for a small, stable area and below 0.33 m for the whole study site (both at 1 σ ).