GM2.3 | From historical images to modern high resolution topography: methods and applications in geosciences
EDI

Recent advances in image collection, e.g. using unoccupied aerial vehicles (UAVs), and topographic measurements, e.g. using terrestrial or airborne LiDAR, are providing an unprecedented insight into landscape and process characterization in geosciences. In parallel, historical data including terrestrial, aerial, and satellite photos as well as historical digital elevation models (DEMs), can extend high-resolution time series and offer exciting potential to distinguish anthropogenic from natural causes of environmental change and to reconstruct the long-term evolution of the surface from local to regional scale.
For both historic and contemporary scenarios, the rise of techniques with ‘structure from motion’ (SfM) processing has democratized data processing and offers a new measurement paradigm to geoscientists. Photogrammetric and remote sensing data are now available on spatial scales from millimetres to kilometres and over durations of single events to lasting time series (e.g. from sub-second to decadal-duration time-lapse), allowing the evaluation of event magnitude and frequency interrelationships.
The session welcomes contributions from a broad range of geoscience disciplines such as geomorphology, cryosphere, volcanology, hydrology, bio-geosciences, and geology, addressing methodological and applied studies. Our goal is to create a diversified and interdisciplinary session to explore the potential, limitations, and challenges of topographic and orthoimage datasets for the reconstruction and interpretation of past and present 2D and 3D changes in different environments and processes. We further encourage contributions describing workflows that optimize data acquisition and processing to guarantee acceptable accuracies and to automate data application (e.g. geomorphic feature detection and tracking), and field-based experimental studies using novel multi-instrument and multi-scale methodologies. This session invites contributions on the state of the art and the latest developments in i) modern photogrammetric and topographic measurements, ii) remote sensing techniques as well as applications, iii) time-series processing and analysis, and iv) modelling and data processing tools, for instance, using machine learning approaches.

Convener: Anette EltnerECSECS | Co-conveners: Livia PiermatteiECSECS, Amaury DehecqECSECS, Katharina AndersECSECS
Orals
| Mon, 24 Apr, 10:45–12:30 (CEST)
 
Room -2.21
Posters on site
| Attendance Mon, 24 Apr, 14:00–15:45 (CEST)
 
Hall X3
Posters virtual
| Attendance Mon, 24 Apr, 14:00–15:45 (CEST)
 
vHall SSP/GM
Orals |
Mon, 10:45
Mon, 14:00
Mon, 14:00

Orals: Mon, 24 Apr | Room -2.21

Chairpersons: Anette Eltner, Livia Piermattei, Katharina Anders
10:45–10:50
10:50–11:00
|
EGU23-6092
|
GM2.3
|
ECS
|
Highlight
|
On-site presentation
Nica Huber, Bronwyn Price, Christian Ginzler, Rolf Holderegger, and Matthias Bürgi

Information regarding the spatial arrangement and extent of habitats in the past is highly important for understanding present biodiversity patterns, assessing restoration potential and fighting extinction-debt effects. Due to increasing intensity of land use, European landscapes have changed profoundly over recent decades, with the trend accelerating following World War 2. Here, we explore the feasibility of deriving a 1946 habitat map for Switzerland compatible and hence comparable with the present-day area-wide habitat map. We take advantage of the newly available orthorectified composite of aerial photographs taken in summer 1946 by the US-Army and provided by swisstopo. The ortho imagery (1 m resolution) is segmented into image objects based on spectral and shape homogeneity for 7 case study areas (320 -508 km2), which represent the main biogeographical regions of Switzerland. Initial training data is derived by manual aerial orthoimage interpretation differentiating 16 habitat classes including wetland, grassland, arable land, hedges, orchard meadows and open forest. A random forest model is trained to classify the segments using variables describing spectral information, image texture, segment shape, topography and climate. To increase the accuracy of the classification, an iterative and semi-automated active learning technique is applied. This technique complements the initial training data with new data for segments with high classification uncertainty. With this contribution, we demonstrate the potential and challenges of object-based image analysis, machine learning and active learning to derive habitat maps from historical black-and-white aerial photography.

How to cite: Huber, N., Price, B., Ginzler, C., Holderegger, R., and Bürgi, M.: Historical habitat mapping based on black-and-white aerial photography, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-6092, https://doi.org/10.5194/egusphere-egu23-6092, 2023.

11:00–11:10
|
EGU23-12274
|
GM2.3
|
On-site presentation
Zuyuan Wang, Yuanyuan Gui, Wei Li, Birgit Eben, Lars T Waser, and Christian Ginzler

Tree-line shifts are evident signs of various aspects of global climatic changes. Remote sensing techniques and aerial imageries are typically used to assess the changes of tree-line in mountainous areas over the past years mostly based on field survey and repeat photography work. The extraction of tree-lines was either done by visual image interpretation or low spatial resolution satellite images. (Bolton et al., 2018)

In Switzerland, the earliest available historical black and white (B&W) images are from the 1920s, and in the other European countries, there is such abundant data as well. Nevertheless, these data sources are currently insufficiently used and there is only limited use of historical aerial images in the analysis of past vegetation. Therefore their automatic and accurate processing still remains challenging. In our previous studies (Wang et al., 2022) covering parts of the Swiss Alps we obtained promising classification results using a deep learning approach. However, difficulties were related to weak and time consuming labeling efforts. In addition, unclear interclass differences between dense forest and group of trees had a negative effect on model accuracies.

In this study, we proposed a BWForest-Unet based on semantic segmentation to access tree cover in Swiss mountain areas. Images from 2019 and labeled tree images using a countrywide canopy height model (CHM) were used in the model. The main advantage of this net is that features from different spatial regions of the image are combined and thus enabling the localization of more precisely regions of interest. The designed BWForest-Unet tries to learn the spatial interdependencies of features by adding an attention model in the decoder processing. Furthermore, suitable data augmentations, e.g., thickness, local elastic, pinch, scratch and grid distortion were applied as an effective method of supplementing the training samples, which intend to efficiently simulate 1980s images by using current 2019 images. The test area consists of 170 1km*1km sample plots distributed over the whole of Switzerland in 1980s.

The study reveals that 1) suitability of semantic segmentation based on BWForest-Unet in combination with B&W aerial images are superior to previous work and therefore promisingto map mountain tree-line change over 35 years in upper tree-line ecotones of the Alps 2) the usese of existing CHMs substantially reduced the labelling workload. 3) The combination of suitable data augmentations simulates the 1980s image to a certain extent.

 

Bolton, D.K., Coops, N.C., Hermosilla, T., Wulder, M.A., White, J.C., 2018. Evidence of vegetation greening at alpine treeline ecotones: three decades of Landsat spectral trends informed by lidar-derived vertical structure. Environmental Research Letters 13, 084022.

Wang, Z., Ginzler, C., Eben, B., Rehush, N., Waser, L.T., 2022. Assessing Changes in Mountain Treeline Ecotones over 30 Years Using CNNs and Historical Aerial Images. Remote Sensing 14, 2135.

How to cite: Wang, Z., Gui, Y., Li, W., Eben, B., Waser, L. T., and Ginzler, C.: Can historical black and white images be used to map changes of tree-line?, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-12274, https://doi.org/10.5194/egusphere-egu23-12274, 2023.

11:10–11:20
|
EGU23-1852
|
GM2.3
|
On-site presentation
Orkan Özcan, Emrah Özpolat, Semih Akay, Okan Özcan, and Tolga Görüm

The Çukurova Delta Complex, which is located in the south of Turkey along the northeastern corner of the Mediterranean Sea, is the second-largest delta system in the Mediterranean. The Seyhan River flowed 10 km east from its current course until at least the 16th Century, and shifted to its current course in the west and began to build the modern delta and the youngest foredune ridges were formed by a combination of aeolian and littoral processes. Morphometrics of foredunes greatly contributes to understanding the relationship between aeolian and marine dynamics. High-resolution digital elevation models (DEMs) are important in examining the geomorphic features of foredune ridges because the low-relief delta environment makes it difficult to use standard topographic maps. Therefore, in the study, the morphometric features, including foredune height, foredune slope, foredune width, the space between foredune ridges, and beach width of the ridges within the study area were extracted from the DEM and orthophotograph of historical and recent aerial photographs. Structure from Motion (SfM) techniques allow for the reconstruction of present and past landforms, and to detect long-term topographic changes in the low-relief areas using historical and modern aerial images. A total of 27 aerial photographs were acquired from flights in AD 2016 covering the study area with a ground sampling distance of 0.3 m, while 13 archive analog aerial photographs with a ground sampling distance of 0.7 m were available from flights in AD 1950. Analysis of SfM-derived high-resolution DEM for the Seyhan Delta shows at least 25 foredune ridges inland for 4 km. It is very important to know the origin and morphodynamics of ridges in terms of revealing the coastal evolution of the Seyhan Delta. Since these ridges preserve past shoreline positions Holocene foredune ridges in the study area can be used to help reconstruct the nature of paleoenvironmental change.

 

How to cite: Özcan, O., Özpolat, E., Akay, S., Özcan, O., and Görüm, T.: Reconstruction of landforms using historical and recent aerial photographs for landscape evolution of coastal dune dynamics, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-1852, https://doi.org/10.5194/egusphere-egu23-1852, 2023.

11:20–11:30
|
EGU23-17437
|
GM2.3
|
On-site presentation
Wilfried Karel

This work is on the automated, photogrammetric orientation of World War II aerial reconnaissance images. For these near-nadir images, the footprint centers are known beforehand with an accuracy of a few hundred meters, together with coarse image scales and nominal focal lengths, but without image rotations. Since their overlap is typically small or absent, the approach orients the images one after another. A novel rotation-invariant, end-to-end CNN image feature matcher finds homologous points both in an aerial image and in an iteratively refined detail of a present-day orthophoto map, initially extracted according to the given metadata. This results in automatically determined ground control points whose heights are interpolated in a likewise present-day terrain model, and which serve to estimate image orientations in a bundle adjustment. The image orientation quality is assessed by projecting manually observed ground control points into image space and comparing them to their likewise manually observed image positions. Hundreds of images of various scales are evaluated, featuring cloud and snow cover, long cast shadows, dust, and scratches. Despite the large gap in time between the aerial and reference data sets and respectively large changes on the ground and in appearance, the approach results in an RMSE of manual ground control points of less than 1mm for over 60% of the images.

How to cite: Karel, W.: Deep Georeferencing of WWII Aerial Reconnaissance Images, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-17437, https://doi.org/10.5194/egusphere-egu23-17437, 2023.

11:30–11:40
|
EGU23-10244
|
GM2.3
|
ECS
|
Virtual presentation
|
Simon Walker, Scott Wilkinson, Rebecca Bartley, Shaun Levick, Anne Kinsey-Henderson, Sana Khan, and Pascal Castellazzi

Accurate measurements of geomorphic change are necessary to improve quantitative and conceptual models in geomorphology. New generation high-resolution topography (HRT) is enabling increasingly accurate quantification of surface change via differencing of fine scale (<1 m) multi-temporal digital elevation models (DEMs). The resulting DEMs of difference (DoDs) provide spatially continuous estimates of surface change. However, harnessing the information contained in HRT DoDs requires progressively sophisticated methods for handling the error propagated into a DoD from each DEM. As HRT acquisition increases, and technology to host and distribute the data improves, there is increasing need for reliable and repeatable error handling procedures. We investigate the potential for satellite-borne optical and radar data to improve DEM-based geomorphic change detection in semi-arid landscapes. The primary motivation for this work is to enable improved geomorphic change detection in semi-arid landscapes affected by extensive erosion. We apply the methodology to a ~15 km2 catchment adjacent to the Great Barrier Reef, Australia, where independent end-of-catchment sediment load data is available for comparison. Our goal is to enable improved geomorphic change detection over relatively large areas (>1 km2) by minimising systematic error propagated into a DoD, particularly from DEMs with sparse ground control networks.

We find the methodology reliably decreases the systematic error in our DoD and improves the separation of real geomorphic change from noise. However, the presence of grass and consequent point misclassification remains a key challenge even with a relatively high point density (~48 pts·m2) airborne-lidar dataset. This is the first time optical and radar remote sensing have been used alongside airborne-lidar for improved DEM-based geomorphic change detection in semi-arid landscapes.

How to cite: Walker, S., Wilkinson, S., Bartley, R., Levick, S., Kinsey-Henderson, A., Khan, S., and Castellazzi, P.: Optical and radar remote sensing improve DEM-based geomorphic change detection in semi-arid landscapes, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-10244, https://doi.org/10.5194/egusphere-egu23-10244, 2023.

11:40–11:50
|
EGU23-252
|
GM2.3
|
ECS
|
Virtual presentation
Kavita Vaijanath Mitkari, Sanjeev Sofat, Manoj K Arora, and Reet Kamal Tiwari

Changes in glacier area, snow, ice, debris cover, and other geomorphological features such as debris cones have a significant impact on the glacial dynamics, are a direct measure of glacier advance and retreat, form a critical input for measuring glacier mass balance, help identify the location of equilibrium line altitude, contribute to the global sea-level rise, and are a good index for understanding local climatic changes. Formation of glacial lakes enhance the rate of glacial melting and catastrophic events arising out of the outburst of glacial lakes can have serious impacts on the human life and economy. So, monitoring the spatial and temporal changes of glacier surface as well as non-surface features is imperative for assessing the health of glaciers and their behavior toward the climate change. The availability of high spatial resolution remote sensing images, has made precise mapping and monitoring of the changes in the glacier surface features and geomorphological features viable at a local level using object-based change detection (OBCD) rather than traditional pixel-based change detection (PBCD). OBCD has been used in numerous applications however, it has received little attention within the glaciological community. Advantage of using OBCD over PBCD is that the object-based paradigm enables the characterization of different land cover classes within the same image, using different object sizes. Further, in OBCD, each image object is considered as a single entity and hence, the small spurious changes and misregistration errors that occur due to high spectral variability are reduced because segmentation generates image objects which are less sensitive to the small spurious changes and misregistration respectively. Furthermore, a comprehensive literature survey on the Gangotri Glacier, Indian Himalayas uncovered that so far, no work has been done linking the variation of glacier surface and non-surface features with the important climate variables that is, temperature and precipitation. Therefore, this study has evaluated the changes in the Gangotri Glacier features at a large scale using class OBCD approach from high spatial resolution WorldView-2 and LISS-4 images for a three-year period from 2011-2014. The meteorological data of Gangotri Glacier was obtained from Climate Research Unit Time Series v.4.06 dataset. A surge in the annual mean temperature and decline in the annual precipitation caused snow/ice area reduction by ~52%. This is accompanied by an increase in the ice-mixed debris (IMD) area by ~11%. The increase in IMD may lead to enhanced ice melting as it could reflect less incoming solar radiations. This further should have revealed expansion in supraglacial debris (SGD) area, however, it has minimized by ~0.4% which is justified with a rise in the periglacial debris (PGD) and debris cones by ~21% and ~9% respectively. Ascend in the annual mean temperature has also shown an increase of ~70% in the area of supraglacial lakes (SGLs), though the number of SGLs decreased; decrease in the number of SGLs suggests widening of SGLs in area. Thus, the dynamics of the glacier features is greatly affected by the yearly temperature and precipitation alterations in the area.

How to cite: Mitkari, K. V., Sofat, S., Arora, M. K., and Tiwari, R. K.: Linking Changes in Gangotri Glacier Features Derived at a Large-Scale with Climate Variability, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-252, https://doi.org/10.5194/egusphere-egu23-252, 2023.

11:50–12:00
|
EGU23-10018
|
GM2.3
|
ECS
|
On-site presentation
|
Francesco Ioli, Francesco Nex, and Livio Pinto

High-frequency monitoring of hardly-accessible glaciers is usually challenging. Though, it is critical for understanding and modeling infra-seasonal glacier dynamics. Fixed time-lapse camera are often used for retrieving high-frequency qualitative and quantitative information on glaciers' evolution. Nevertheless, only one camera is usually employed for estimating glacier surface velocity by Digital Image Correlation (DIC) techniques and an approximate DSM is required [1]. Using multiple cameras can step up in-situ glacial monitoring, as 3D scene reconstruction can be obtained by photogrammetry and Structure-from-Motion (SfM). Indeed, two (or more) cameras allows for estimating glacier surface flow velocity in a 3D world, volume variations, ablation and glacier terminus retreat.


This work presents a pilot study for implementing a low-cost image-based stereoscopic system for automatic high-frequency monitoring of an alpine glacier. Each hand-made monitoring station includes a DSLR camera, an Arduino microcontroller for camera triggering, a Raspberry Pi Zero with a SIM card for sending images to a remote server. The two cameras were installed in summer 2021 on each side of the Belvedere Glacier north-west terminus (Italian Alps), with a wide baseline (i.e., ∼260 m). The cameras have been operating taking daily images for one and a half year. Every day, the acquired stereo-pair was processed by SfM. Due to the wide baseline, which is typical of complex mountain environments, finding corresponding points across different viewpoints was troublesome [2]. Commercial SfM software packages based on traditional feature matching (e.g., Agisoft Metashape) failed to find enough and well distributed matches, while state-of-the-art deep learning-based algorithms for wide-baseline matching, such as SuperGlue [3], outperformed traditional feature matching. Therefore, an automatic open-source Python pipeline for finding matches, orienting image-pairs, solving Bundle Adjustment with Ground Control Points (GCPs) and building 3D point clouds was developed from scratch. Although alternative open-source solutions are under study, dense 3D reconstruction is currently carried out at every epoch by Agisoft Metashape, exploiting Python API to fully integrate dense matching in the processing pipeline. Results were validated at three epochs by UAV-based ground truth, obtaining RMSE of point clouds of ∼15 cm.

Overall, the monitoring system is simple and low-cost (less than €2000 per camera), requires minimum in-situ operations (limited to cameras’ installation and materialization of few GCPs), and an automatic 3D processing of stereo-pairs can improve in-situ glacier monitoring. Indeed, from daily point clouds, glacier volume reduction and retreat speed can be estimated by computing cloud-to-cloud distances. This, combined with surface velocities estimated by DIC, may help glaciologists to better understand glacier dynamics and quantify mass balances. The full Python pipeline will be released as open-source code, together with a documentation to make it reproducible for other study cases.

[1] Messerli, A., & Grinsted, A. (2015). Image georectification and feature tracking toolbox: ImGRAFT. Geosci. Instrum. Meth., 4(1), 23–34.

[2] Yao, G., Yilmaz, A., Meng, F., & Zhang, L. (2021). Review of Wide-Baseline Stereo Image Matching Based on Deep Learning. Remote Sens., 13(16)

[3] Sarlin, P. E., Detone, D., Malisiewicz, T., & Rabinovich, A. (2020). SuperGlue: Learning Feature Matching with Graph Neural Networks. Proc. CVPR IEEE, 4937–4946.

How to cite: Ioli, F., Nex, F., and Pinto, L.: High-frequency automatic 3D glacier monitoring using low-cost time-lapse cameras and Deep Learning algorithms, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-10018, https://doi.org/10.5194/egusphere-egu23-10018, 2023.

12:00–12:10
|
EGU23-13783
|
GM2.3
|
ECS
|
On-site presentation
Niccolò Dematteis, Fabrizio Troilo, Riccardo Scotti, Davide Colombarolli, Daniele Giordan, and Valter Maggi

Glacier surface velocity is a crucial information to assess the impact of global warming on glaciers, since it is related to the ice thickness; its variations provide information on mass balance and, in general, on the current state of glacier “health”. Moreover, velocity anomalies are often an indicator of glacier instabilities. Therefore, attention has been dedicated to surveying glacier velocity. Historically, surface velocity was the first quantitative variable measured on glaciers since the 19th century, using phototheodolites. In the last decades, terrestrial monoscopic digital time-lapse cameras (TLC) have permitted to conduct automatic surveying for long periods at high spatial and temporal resolutions using digital image correlation. Even though terrestrial time-lapse imagery is currently a consolidated technique in glacier monitoring, the number of dedicated publications is relatively small. In particular, possible strategies, limitations and potentialities have never been systematically reviewed.

This work aims to illustrate the typical procedures required to monitor glacier surface velocity using terrestrial monoscopic TLC, which can be synthetically listed as: 1) correct deployment of the equipment and image acquisition; 2) data pre-processing: 2.1) image selection, 2.2) colour/feature enhancement and 2.3) image registration; 3) data processing: displacement measurement using image correlation; 4) data post-processing: 4.1) outlier correction, 4.2) image geocoding and 4.3) time-series extraction. We describe possible inconveniences that can arise during the survey – e.g., image misregistration, distortion and defocusing, illumination and chromatic variation (shadows, snow patches), presence of outliers, and geocoding issues – and provide some guide lines to minimise such problematics. We present six study cases in the European Alps – Planpincieux, Grandes Jorasses, Freney and Brenva glaciers in the Mont Blanc massif, and Western and Eastern Fellaria glaciers in the Bernina massif – that feature different monitoring equipment, site geometry and glacier morphodynamics to illustrate possible solutions for terrestrial imagery monitoring.

The results revealed that terrestrial TLC provided high spatial resolution and acquisition frequency to detect small kinematic sectors and fast-occurring velocity anomalies, which would be difficult to identify using alternative approaches (e.g., satellites or topographic). However, like other passive optical sensors, the principal limitation is that they are affected by poor visibility and cannot acquire during the night. This study highlighted the great potentiality of TLC in glacier kinematics surveying, which can be conducted using either professional cameras or low-cost webcam and IP cameras, according to the scope and financial availability. The contained costs and ease of installation make TLC a very high benefit-to-cost tool and permit the development of strategies for widespread glacier monitoring at a regional scale with relatively low financial efforts.

How to cite: Dematteis, N., Troilo, F., Scotti, R., Colombarolli, D., Giordan, D., and Maggi, V.: Monoscopic terrestrial time-lapse cameras: an effective tool for surveying glacier surface velocity, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-13783, https://doi.org/10.5194/egusphere-egu23-13783, 2023.

12:10–12:20
|
EGU23-6469
|
GM2.3
|
ECS
|
On-site presentation
Sebastian Mikolka-Flöry, Camillo Ressl, and Norbert Pfeifer

Historical images are an important resource for documenting the early states of our environment after the last little ice age. To extract a feature (e.g. glacier outline) from a single historical oblique image in a global coordinate system monoplotting is commonly used: Rays originating from the projection center passing through the pixel vertices, which represent the considered feature, in the image are intersected with a reference terrain model. A subsequent spatial analysis not only requires the 3D position of these vertices as result of monoplotting but also their positional accuracy. The derivation of the latter has not been properly addressed so far.

Existing approaches for assessing the monoplotting accuracy are either based on i) reference data or ii) selected ground control points (GCPs). The first approach is generally not suitable for historical images as reference data is mostly not available. Evaluation based on GCPs is only a rough measure for the potentially achievable accuracy as the monoplotting accuracy varies strongly within an image and the number of GCPs is usually limited. 

Hence, we propose a new approach based on variance propagation. Formulating the monoplotting principle using projective geometry both the accuracy of the estimated camera parameters as well as the reference terrain are considered within the estimation of the uncertainty for the 3D position of each vertex. Estimating the uncertainty for each vertex of the monoplotted feature further allows to derive a differentiated analysis of the results. Furthermore, being independent from necessary reference data our approach is well suited for historical images. Hence, with the developed approach it becomes possible to consider the uncertainty of monoplotted features in subsequent spatial analyses which is especially important when comparing these features with modern reference datasets; e.g. in order to judge the significance of possible changes or deformations.

How to cite: Mikolka-Flöry, S., Ressl, C., and Pfeifer, N.: Uncertainty of monoplotted features from historical single oblique images, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-6469, https://doi.org/10.5194/egusphere-egu23-6469, 2023.

12:20–12:30
|
EGU23-6554
|
GM2.3
|
ECS
|
On-site presentation
Thomas Goelles, Birgit Schlager, and Stefan Muckenhuber

Point clouds can be acquired by different sensor types and methods, such as lidar (light detection and ranging), radar (radio detection and ranging), RGB-D (red, green, blue, depth) cameras, SfM (structure from motion), etc. 

In many cases multiple point clouds are recorded over time, sometimes also referred to as 4D point clouds. For example, automotive lidars from Ouster or Velodyne record point clouds at around 10-20Hz resulting in millions of points per second. In addition, monitoring with terrestrial laser scanners is becoming used more often. Producing similar datasets than the automotive lidars, although with larger individual point clouds at a lower frame rate.

Analyzing such a large collection of point clouds is a big challenge due the size and unstructured nature of the data. The Python package "pointcloudset" provides a way to store, analyze, and visualize large datasets consisting of multiple point clouds recorded over time. Pointcloudset features lazy evaluation, parallel processing and is designed to enable the development of new point cloud algorithms and their application on big datasets. The package is based on the Python packages pandas, pytncloud, dask and open3D. Its API is easy to use and high level and the package is open source and available on GitHub. 

How to cite: Goelles, T., Schlager, B., and Muckenhuber, S.: pointcloudset - A Python package to analyze large datasets of point clouds recorded over time, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-6554, https://doi.org/10.5194/egusphere-egu23-6554, 2023.

Posters on site: Mon, 24 Apr, 14:00–15:45 | Hall X3

Chairpersons: Livia Piermattei, Katharina Anders, Anette Eltner
X3.18
|
EGU23-2715
|
GM2.3
|
ECS
Moritz Altmann, Florian Haas, Jakob Rom, Fabian Fleischer, Tobias Heckmann, Camillo Ressl, and Michael Becht

Since the end of the Little Ice Age around 1850, global warming has led to rapid landscape changes, especially in high mountain areas. The ongoing glacier melt leads to an expansion of the LIA glacier forefields, so-called proglacial areas. The exposed lateral moraines often show increased sediment activity over decades and centuries, which is generally described as the paraglacial adjustment process. Slope instabilities are caused, for example, by the loss of the support from the melting glaciers, which can lead to large landslides and thus heavy deformations. In order to understand corresponding geomorphological processes, it is important that surface changes can be reconstructed and analysed in high spatial and temporal resolution. However, aerial photographs of the European Alps, which are well suited for observing proglacial areas, only extend to the middle of the 20th century, thus resulting in a temporal limitation.

Therefore, in this work we show a nearly 100-year quantitative monitoring of a large-scale deformation of a LIA lateral moraine in the glacier forefield of the Gepatschferner in the upper Kaunertal (Tyrol, Austria). We achieve this long-term (1922 to 2021) observation by combining different topographic data sets based on different remote sensing methods and techniques. The reconstructed earth surfaces are based on airborne LiDAR data (2006 to 2021) and photogrammetric DEMs (1953 to 2003) as well as a historical stereophotogrammetric map from 1922, which was also generated into a DEM. In total, eight DEMs were generated and corresponding DoDs calculated.

Different landslides within the first three epochs (1922 to 1953, 1953 to 1971 and 1971 to 1983) could be determined on the slope, which can be directly linked to the corresponding glacier melt. Even after the landslide processes (from 1983 onwards), continuous geomorphological activity could be observed until today (2021), whereby the total volume of net erosion of all epochs (from 1922 to 2021) added up to approx. 486,000 m³.

How to cite: Altmann, M., Haas, F., Rom, J., Fleischer, F., Heckmann, T., Ressl, C., and Becht, M.: Long-term reconstruction of a large-scale landslide of a LIA  lateral moraine in the Upper Kaunertal in Tyrol, Austria, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-2715, https://doi.org/10.5194/egusphere-egu23-2715, 2023.

X3.19
|
EGU23-13315
|
GM2.3
Mateja Breg Valjavec, Rok Ciglič, Matjaž Geršič, and Špela Čonč

Dolines are concave circular karst landforms that are clearly presented by topography data (topographic maps), especially on high-resolution LiDAR digital terrain models (DTM). In the karst landscape, man has reshaped natural dolines through centuries by collecting rocks and soil to increase the flat area of tillable (cultivated) land at the bottom of the doline. By this human-induced process, the natural doline was reshaped into a cultivated doline. Cultivated dolines have a rich historical legacy of use for local agricultural production, high geomorphological value for geodiversity and present an important habitat supporting biodiversity. They are an element of agro-karstic landscape (paysage agro-karstiques) and are distinctive of Mediterranean karst landscapes like Dinaric karst, Central massive, Apulia etc. Most of the cultivated dolines have been recently abandoned, and covered by forest, thus the human impact is not evident anymore so clearly.

In most studies on natural characteristics and processes in dolines, there is no distinct separation between natural and cultivated dolines and no consideration of past agricultural land use. Thus, the main goal of this study/presentation is to provide a geoinformatics methodology to separate cultivated dolines from natural dolines based on differences in micro-topography by using recent very high-resolution LiDAR topography data and historical cadastral maps (19th century).

Using visualized LiDAR DTM the most evident morphometric differences between the natural and cultivated doline landforms were recognized. Cultivated dolines were characterized by (1) a circular concave landform with a flat bottom (2) the presence of anthropogenic elements, such as circular stonewalls at the upper doline edge, which provides evidence of stone-removal from the doline slopes (smooth surface). Additionally, in the 19th-century cadastral maps, only individual dolines with important land use were marked as special lots. Given the rural character of the landscape during that time, the only land use recorded in the concentric lots of the dolines was agricultural use (arable fields, gardens, meadows, and pastures). As a result, the number, location and surface coverage of cultivated dolines were precisely defined for classical karst regions in SW Slovenia. Based on Lidar data, bowl-shaped cultivated dolines with flat bottoms were separated from non-cultivated funnel-shaped dolines.

 

How to cite: Breg Valjavec, M., Ciglič, R., Geršič, M., and Čonč, Š.: Application of historical cadastral maps and high resolution airborne LiDAR topography to distinguish anthropogenic from natural karst landforms: case study of karst dolines, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-13315, https://doi.org/10.5194/egusphere-egu23-13315, 2023.

X3.20
|
EGU23-7007
|
GM2.3
|
ECS
Xabier Blanch, Anette Eltner, Marta Guinau, and Antonio Abellán

In recent years, photogrammetric models have gained widespread use in geosciences due to their ability to reproduce natural surfaces. These models offer a cost-effective and user-friendly alternative to other systems, such as LiDAR, for creating 3D point clouds. On the other hand, rockfalls pose a significant risk to society, as they are the most common natural hazard in mountainous areas and can occur with great speed, resulting in high levels of danger. The aim of this communication is to show results on the development of new algorithms and time-lapse photogrammetric systems for automatic rockfall monitoring (Blanch, 2022).

To acquire the data, a photogrammetric system consisting of different photographic modules and a data transmission module has been developed. This system uses conventional cameras (24Mpx-48Mpx) powered by solar panels and it is controlled by a Raspberry Pi. The system captures time-lapse images, can be programmed, configured flexibly, and it can send images remotely for near real-time processing. The system has been installed at two sites with rockfall activity. One in the Puigcercós cliff, located in the Origens UNESCO Gobal Geopark (Spain), and the other in the Tajo de San Pedro cliff located in the Alhambra de Granada - UNESCO World Heritage Site (Spain).

Data processing comprises two main steps. The first step involves the automatic photogrammetric process using SfM-MVS algorithms. Thereby, the MEMI workflow is applied to improve the level of detection in the change-detection comparison (Blanch et al., 2021). Afterwards, a workflow based on M3C2 (Lague et al., 2013) comparison and DBSCAN clustering is applied to identify possible rockfall clusters. The resulting clusters are processed via a machine learning approach to automatically discriminate the true rockfall events from the candidate clusters . To perform this task, various metric parameters, i.e. features, of the candidate clusters are calculated, and a Random Forest machine learning model is used to perform the classification.

The presented approach facilitates the automated monitoring of rockfalls in near-real time, while improving the detection threshold in the 3D change-detection models, resulting in a more detailed characterisation of active zones and defining the framework that allows for automated 4D rockfall monitoring in high temporal frequency.

Blanch, X., 2022.  Developing Advanced Photogrammetric Methods for Automated Rockfall Monitoring. Doctoral dissertation. http://hdl.handle.net/10803/675397

Blanch, X.; Eltner, A.; Guinau, M.; Abellan, A., 2021. Multi-Epoch and Multi-Imagery (MEMI) Photogrammetric Workflow for Enhanced Change Detection Using Time-Lapse Cameras. Remote Sens. , 13, 1460. https://doi.org/10.3390/rs13081460

Lague, D., Brodu, N., Leroux, J., 2013. Accurate 3D comparison of complex topography with terrestrial laser scanner: Application to the Rangitikei canyon (N-Z). ISPRS Journal of Photogrammetry and Remote Sensing 82, 10–26. https://doi.org/10.1016/j.isprsjprs.2013.04.009

How to cite: Blanch, X., Eltner, A., Guinau, M., and Abellán, A.: Photogrammetric time-lapse workflow for automated rockfall monitoring, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-7007, https://doi.org/10.5194/egusphere-egu23-7007, 2023.

X3.21
|
EGU23-8158
|
GM2.3
|
ECS
Katharina Anders and Bernhard Höfle

Surface processes in topographic data are typically extracted either as local surface changes with static spatial extent between multiple acquisitions, or by tracking features or objects with more or less rigid properties which are re-identified in each epoch. Such approaches are challenged when surface processes are highly dynamic, such as material transport of sand or snow as moving and deforming forms. For the observation of dynamic surface processes, strategies of near-continuous 3D acquisition, e.g. permanent laser scanning or time-lapse photogrammetry, capture dense time series of point clouds of a scene. To extract surface processes as moving spatiotemporal objects from these datasets, we propose a time-extended approach to the extraction of 4D objects-by-change [1]. These objects are automatically identified in their spatial and temporal extent in 3D time series by first detecting surface activities in the time series at a location, and then spatially delineating them based on similar change histories (i.e. surface behavior in time) throughout their duration. So far, this method was temporally static, meaning that the timing and duration was fixed for each 4D object-by-change. By extending the search for similar change histories along the time domain, we enable to trace a moving object through the space-time coverage of a dataset. We demonstrate the method using simulated 3D time series and present first results for real-world near-continuous 3D data of sediment transport. The method will be openly accessible in py4dgeo [2], an open source Python library for change analysis in 4D point clouds. Advantages over existing methods are that no a-priori information on specific processes are required, and no definition of distinct features to be tracked is needed. A major strength is the novel possibility to delineate surface processes as intangible objects in space and time, which holds potential to provide completely new information on surface dynamics in topographic scenes.

 

References:

[1] https://doi.org/10.1016/j.isprsjprs.2021.01.015

[2] https://github.com/3dgeo-heidelberg/py4dgeo

How to cite: Anders, K. and Höfle, B.: Spatiotemporal tracking of surface processes through their change histories in dense 3D time series by implementing a time-extension on the 4D objects-by-change method, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-8158, https://doi.org/10.5194/egusphere-egu23-8158, 2023.

X3.22
|
EGU23-7115
|
GM2.3
|
ECS
|
Mathilde Letard, Dimitri Lague, Arthur Le Guennec, Sébastien Lefevre, Baptiste Feldmann, Paul Leroy, Daniel Girardeau-Montaut, and Thomas Corpetti

Three-dimensional data have become increasingly present in earth observation over the last decades and, more recently, with the development of accessible 3D sensing technologies. However, many 3D surveys are still underexploited due to the lack of accessible and explainable automatic classification methods. In this work, we introduce explainable machine learning for 3D data classification using Multiple Attributes, Scales, and Clouds under 3DMASC, a new workflow. It handles multiple clouds at once, including or not spectral and multiple returns attributes. Through 3DMASC, we use classical 3D data multi-scale descriptors and new ones based on the spatial variations of geometrical, spectral and height-based features of the local point cloud. We also introduce dual-cloud features, encrypting local spectral and geometrical ratios and differences, which improve the interpretation of multi-cloud surveys. 3DMASC thus offers new possibilities for point cloud classification, namely for the interpretation of bi-spectral lidar data. Here, we experiment on topo-bathymetric lidar data, which are acquired using two lasers at infrared and green wavelengths, and feature two irregular point clouds characterized by different samplings of vegetated and flooded areas, that 3DMASC can harvest. By exploring the contributions of 88 features and 30 scales – including two types of neighborhoods – we identify a core set of features and scales particularly relevant for coastal and riverine scenes description, and give indications on how to build an optimal predictor vector to train 3D data classifiers. Our findings highlight the predominance of lidar return-based attributes over classical features based on dimensionality or eigenvalues, and the significant contribution of spectral information to the detection of more than a dozen of land and sea covers – artificial/vegetated/rocky/bare ground, rocky/sandy seabed, intermediate/high vegetation, buildings, vehicles, power lines. The experimental results show that 3DMASC competes with state-of-the-art methods in terms of classification performances while demanding lower complexity and thus remaining accessible to non-specialist users. Relying on a random forest algorithm, it generalizes and applies quickly to large datasets, and offers the possibility to filter out misclassified points depending on their prediction confidence. Classification accuracies between 91% for complex scene classifications and 98% for lower-level processing are observed, with average prediction confidences above 90% and models relying on less than 2000 samples per class and at most 30 descriptors – including both features and scales. Though dual-cloud features systematically outperform their single cloud equivalents, 3DMASC also performs on single cloud lidar data, or structure from motion point clouds. Our contributions are made available through a self-contained plugin in CloudCompare allowing non-specialist users to create a classifier and apply it, and an opensource labelled dataset of topo-bathymetric data.

How to cite: Letard, M., Lague, D., Le Guennec, A., Lefevre, S., Feldmann, B., Leroy, P., Girardeau-Montaut, D., and Corpetti, T.: 3DMASC: accessible, explainable 3D point clouds classification. Application to bi-spectral topo-bathymetric LiDAR data., EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-7115, https://doi.org/10.5194/egusphere-egu23-7115, 2023.

X3.23
|
EGU23-10162
|
GM2.3
|
ECS
|
Highlight
|
Gerardo Zegers, Alex Garces, and Masaki Hayashi

Accurate estimation of surficial sediment size in alpine landforms such as talus slopes, rock glaciers, and moraines is crucial for understanding geomorphologic processes and predicting the potential impact of natural hazards. Traditional methods for measuring sediment size in these environments can be time-consuming and labor-intensive. Additionally, they are usually applied to selected areas and are rarely used to cover larger areas, making the development of more efficient approaches essential. This study presents a new method for estimating large-scale surficial sediment size based on unmanned aerial vehicle (UAV) photogrammetry and combining SediNet and PebbleCountAuto image analysis methods. SediNet is a configurable machine-learning framework for estimating either (or both) continuous and categorical variables from a photographic image of clastic sediment. SediNet can achieve subpixel resolutions because the dimensions of the grains aren't being measured directly. However, site-specific sediment sizes are necessary to train this model. On the other hand, PebbleCountAuto does not require any site calibration by using segmentation methods to delimitate the grains automatically and provide a full grain-size distribution. Our proposed methodology trains the SediNet model using the sediment sizes outputs of the PebbleCountAuto method. Our study area is the upper part of the Lake O'Hara watershed in the Canadian Rockies, composed of talus slopes and a large ice-cored moraine. We performed two types of UAV flights; high-altitude flights (~100 m height) to cover the whole study area with medium-to-high resolution orthomosaic (pixel resolution 3 cm) and low-altitude flights (~25 m height) at smaller patches to achieve high-resolution orthomosaic (pixel resolution 5-8 mm). First, the sediment size was estimated in the high-resolution patches with the PebbleCountAuto method. Then, these results were used to train the SediNet model and generate a large-scale sediment size estimation. Our results show that this combination of methods is a reliable and efficient approach for accurately estimating sediment sizes in alpine landforms. The use of UAV photogrammetry allows for the rapid and cost-effective collection of high-resolution imagery, while the combination of SediNet and PebbleCountAuto provides robust estimates of sediment size over large areas. This new method can improve our understanding of geomorphologic processes and hazard assessment in these environments.

How to cite: Zegers, G., Garces, A., and Hayashi, M.: Large-scale estimation of surficial sediment size in alpine landforms using UAV photogrammetry and machine learning., EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-10162, https://doi.org/10.5194/egusphere-egu23-10162, 2023.

X3.24
|
EGU23-13606
|
GM2.3
|
ECS
Špela Čonč and Mateja Breg Valjavec

Karst is a geomorphological system that covers almost 50% of the area of Slovenia and is mainly characterised by circular concave and convex landforms such as conical hills, dolines, uvalas and poljes. These large landforms can be easily detected with a number of already developed (semi-)automatic detection methods. In addition to these large landforms, the karst surface is dissected by smaller scale features consisting mainly of numerous rocks of different shapes and sizes. Due to the different lithology that makes up the Slovenian karst (e.g., limestone, dolomite), rocky outcrops have different morphographic and morphometric characteristics due to the different dynamics of the mechanical weathering of the bedrock. The variety in shapes and sizes of rocky outcrops makes their detection by automatic or semi-automatic methods difficult. In our study area in the Dinaric Karst in Slovenia, they reach heights of up to several metres and lengths of about 10 metres.

Field mapping or digitizing such landforms would be time-consuming, labour-intensive, and costly. The combination of high-resolution LiDAR-derived DEMs (digital elevation models) and (semi-)automatic landform detection and delineation methods in GIS environments enables remote and low-cost mapping, which has an outstanding potential for large-scale spatial analysis and mapping in remote, forested, and difficult-to-access areas such as the Dinaric Karst.

The main objective of this study was to develop an approach for quantitative identification and detection of rocky outcrops. The approach is based on spatial analysis of high-resolution (1 m × 1 m) LiDAR DEM and field analysis of outcrop morphography and morphometry. The study was conducted in the Dinaric Karst area in Slovenia, which consists mainly of Cretaceous and Jurassic limestones and dolomites. First, we calculated the values of TPI (Topographic Position Index) to identify all convex shapes (i.e., ridges) within a search radius of 10 m around each cell. Slope was used as an additional criterion for defining rocky outcrops. Based on field measurements, we found that bedrock in areas with limestone (30°) outcrop at a lower surface slope than in areas with dolomite (50°).

The study has shown that the different spatial distribution, shape and size of the rocky outcrops are related to the geological structure. In the limestones they are much denser and more numerous than in the dolomites. In average, the dimensions of the outcrops are also much larger. This is due to the porosity of the dolomites, which causes greater mechanical weathering. We have also found that rocky outcrops often occur on certain landforms, e.g. on the slopes of dolines and other karst depressions or fluviokarst valleys.

Airborne LiDAR DEMs can be a useful source of information for detecting and studying the spatial patterns and morphometric settings of rocky outcrops. The number of landforms detected indicates that, in addition to dolines, rocky outcrops are one of the most common landforms in the Dinaric Karst.

Key words: Rocky outcrops, karst landform, GIS, LiDAR, (semi-)automatic methods, geomorphology, Dinaric mountains

How to cite: Čonč, Š. and Breg Valjavec, M.: Detection of rocky outcrops from LiDAR-derived DEM in Dinaric Karst, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-13606, https://doi.org/10.5194/egusphere-egu23-13606, 2023.

X3.25
|
EGU23-13008
|
GM2.3
|
ECS
Robert Krüger, Xabier Blanch Gorriz, Oliver Grothum, and Anette Eltner

Between June and August 2022, the European Forest Fire Information System (EFFIS) reported more fires in Europe than in any other recent summer season. This is particularly true for Central Europe, where the largest forest fire in recent Czech history occurred in the German-Czech border region. With global warming and resulting longer dry periods, the length and severity of wildfire seasons in central Europe will likely increase. Therefore, easy to implement and cost-effective methods to assess wildfire damage and regeneration of the ecosystems are getting increasingly important. In this study we evaluated how different datasets obtained by uncrewed aerial system (UAS) can be incorporated with datasets obtained from the ground to describe the fire affected landscape. Thereby, multi-spectral 3D point clouds were derived from low-cost UAV laser scanning and using structure from motion (SfM) photogrammetry applied to RGB and multi-spectral imagery. The aerial datasets were combined with ground-based terrestrial and mobile laser scanning. The datasets were acquired in several surveys following the forest fire event in the German part of the National park Saxonian/Bohemian Switzerland.

Initial results show the potential of UAS-based sensing for efficient mapping of a burned area with a high resolution (600-1000 pts/m²). The combination of point clouds from UAS-based laser scanning and photogrammetry enables a detailed representation of the burned forest with different levels of fire damage (e.g., in still present canopy) when compared to the single datasets. The UAS based laser scanning data reveals higher noise compared to the SfM-based point clouds. However, the accuracy is still sufficient to improve the quality of orthomosaics in densely vegetated areas. In a next step, further investigations on data accuracy are conducted and automated point cloud fusion algorithms based on classified points are considered.

How to cite: Krüger, R., Blanch Gorriz, X., Grothum, O., and Eltner, A.: Using multi-scale and multi-model datasets for post-event assessment of wildfires, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-13008, https://doi.org/10.5194/egusphere-egu23-13008, 2023.

Posters virtual: Mon, 24 Apr, 14:00–15:45 | vHall SSP/GM

Chairpersons: Katharina Anders, Anette Eltner, Livia Piermattei
vSG.2
|
EGU23-7311
|
GM2.3
|
ECS
Hongen Wang

The digital elevation model (DEM) is an important basic data tool applied in geoscience applications. Because of its high cost and long development cycle of enhancing hardware performance, designing the related models and algorithms to improve the resolution of DEM is of considerable significance. At present, Neural networks (NNs) have demonstrated the potential to recover finer textural details from lower-resolution images by super-resolution (SR). Given similar grid-based data structures, some researchers have transferred image SR methods to DEM. These efforts have yielded better results than traditional spatial interpolation methods. However, the deep learning(DL) models need a lot of training data, and the model is difficult to converge, resulting in high training costs, which can be challenging. Therefore, in order to reduce the difficulty and cost of DL method training, we detrend the DEM data to decompose the target DEM into a deterministic low frequency trend part and a high frequency residual part. In the process of DL training, focus on the high-frequency part. We use multiple DL models and DEM data of various landforms to verify, and the experimental results show that our proposed method can indeed reduce the difficulty and cost of DL training. At the same time, our method can also be extended to other DL models.

How to cite: Wang, H.: Super-resolution of digital elevation models using deep learning methods based on detrending, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-7311, https://doi.org/10.5194/egusphere-egu23-7311, 2023.