GM2.11

Mountain glaciers have lost significant mass over the past century in response to a globally warming climate. However, on interannual to decadal time scales, many glaciers in Western North America show periods of both advance and retreat. To better understand these systems and their sensitivity to climate forcing, we are generating regional records of glacier surface elevation change from scanned historical film photographs acquired between the 1950s to 1990s. Our results will help constrain projections of future glacier change under different climate scenarios, as well as impacts on downstream water resources and geohazard risk.

Historical image pre-processing and manual ground control point (GCP) selection are time-intensive bottlenecks during traditional SfM processing workflows. We developed an automated photogrammetry processing pipeline (HSfM) to systematically process large archives of vertical aerial film photographs and generate sub-meter resolution digital elevation models (DEMs), without manual GCP selection. We present several case studies for glaciers in the Western North America using photos from the USGS North American Glacier Aerial Photography (NAGAP) and Earth Explorer Aerial Photography Single Frame archives, which differ in terms of available image overlap, survey area extent, and terrain characteristics. Absolute vertical accuracy of <0.5-1.0 m is achieved through iterative closest point (ICP) co-registration over stable bare-ground surfaces between the historical DEMs and modern high-resolution satellite or lidar reference DEMs. We demonstrate the potential for these new DEM records to quantify geodetic glacier mass balance and geomorphological change including moraine deposition, moraine degradation, and sediment redistribution in proglacial areas.

How to cite: Knuth, F., Shean, D., and Bhushan, S.: Historical Structure From Motion (HSfM): Automated production of high-resolution DEMs from historical aerial photography for long-term geodetic change analysis, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-13196, https://doi.org/10.5194/egusphere-egu21-13196, 2021.

Geographical Surveys now distribute online their historical aerial photographs. The batches of digital images, holding the appearance and relief of the forever gone landscape, can be processed with automated Structure-from-Motion (SFM) photogrammetric pipelines. Are the results trustworthy? In this communication, we report the results of exploratory tests performed with Agisoft Metashape on sets of 1978, ~1/27.000, vertical aerial photographs from IGN-France over la Réunion volcanic island in the Indian Ocean. Georeferencing deliberately used ground control points and check points collected on IGN's web mapping portal. Validation was obtained from lidar and photogrammetric acquisition of 2015.

First, our results show that scanned photographs do not strictly map camera coordinates to image coordinates from one file to the next. Photos are slightly shifted and rotated on each scan. The photogrammetric assumption of a single camera per batch of images is thus violated. A preprocessing step, automated with Python, locates fiducials, computes camera principal point, rotates and crops the image file to a unique image reference frame. This feature is absent from Agisoft Metashape when fiducial coordinates are unknown.

Second, in the photogrammetric pipeline, camera calibration parameters are deduced from matched sparse points. The sensitivity of the "align" function was explored. The smallest RMS errors were ±7.03m for 11 ground-control points and ±5.45m for 9 independent check points when setting Align quality to "high" and a 4-parameters camera model using focal length (f), eccentricity (cx, cy), one radial distortion parameter (K1). A higher number of parameters delivered no accuracy improvement and correlated parameters. Intensive random sampling of sparse points subsets conducted to stable estimates of focal length and eccentricity. Improving the robustness of focal length determination would require additional, oblique photographs, which was not the spirit of historical survey design and were never acquired in past surveys.

Third, collecting ground control points on https://geoportail.gouv.fr resulted in digital surface model elevation accuracy within +/- 3.34m (Median Absolute Deviation). Validation was computed on a 2015 lidar digital terrain model at 5m resolution on stable grounds. Scanning artefacts, probably due to variable scanning velocity of the digitizing head, introduced elevation variation stripes in Difference of DEM (DoD), parallel to the scanner direction. This pattern limits the detection of geomorphologically meaningful differences.

Fourth, a DoD between 2015-1978 for the Cirque de Salazie, in the north-east of La Réunion Island, highlighted landsliding masses active some time during the last 37 years and 13 cyclones. Beyond this proof of concept, archive aerial photographs in La Réunion go back until 1949 and covered the island twenty times. This time scale offers a welcome hindsight when producing landslide risk mitigation maps.

This work was published in open-access in

Rault, C., Dewez, T. J. B., and Aunay, B., 2020, Structure-from-Motion processing of aerial archive photographs: sensitivity analyses pave the way for quantifying geomorphological changes since 1978 in la Réunion island, ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci., V-2-2020, 773–780, https://doi.org/10.5194/isprs-annals-V-2-2020-773-2020, 2020.

How to cite: Dewez, T. J., Rault, C., and Aunay, B.: Structure-from-Motion applied to historical aerial photographs: parameter variability and application on landslide cyclonic evolution on France's La Réunion Island, Indian Ocean, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-2791, https://doi.org/10.5194/egusphere-egu21-2791, 2021.

Glacial and periglacial environments are highly sensitive to climate change, even more in mountain areas where warming is faster and, as a consequence, perennial features of the cryosphere like glaciers and permafrost have been fast evolving in the last decades. In the European Alps, glaciers retreat and permafrost thawing have led to the destabilization of mountain slopes, threatening human infrastructures and inhabitants. The observation of such changes at decadal scales is often limited to sparse in situ observations.

Here, we present three study cases of mountain permafrost sites based on a multidisciplinary approach over almost seven decades. The goal is to investigate and quantify morphodynamic changes and understand the causes of these evolutions. We used stereo-photogrammetry techniques to generate orthophotos and (DEMs) from historical aerial images (available, in France since 1940s). From this, we produced diachronic comparison of DEMs to quantify vertical surface changes, as well as feature tracking techniques of multi-temporal digital orthophotos for estimating horizontal displacement rates. Locally, high-resolution datasets (i.e. LiDAR surveys, UAV acquisitions and Pléiades stereo imagery) were also exploited to improve the quality of photogrammetric products. In addition, we combine these results with geophysics (ERT and GPR) to estimate the ice content, geomorphological surveys to describe the complex environments and the relationship with climatic forcing.

The first study case is the Laurichard rock glacier, where we were able to quantify changes of emergence velocities, fluxes, and volume. Together with an acceleration of surface velocity, important surface lowering have been found over the period 1952-2019, with a striking spatiotemporal reversal of volume balance.

The second study site is the Tignes glacial and periglacial complex, where the changes of thermokarstic lakes surface were quantified. The results suggest that drainage probably affects the presence and the evolution of the largest thermorkarst. Here too, a significant ice loss was found on the central channel concomitant to an increase in surface velocities.

The third study site is the Chauvet glacial and periglacial complex where several historical outburst floods are recorded during the 20th century, likely related to the permafrost degradation, the presence of thermokarstic lakes, and an intra-glacial channel. The lateral convergence of ice flow, due to the terrain subsidence caused by the intense melting, may cause the closure of the channel with a subsequent refill of the thermokarstic depression and finally a new catastrophic event.

Our results highlight the important value of historical aerial photography for having a longer perspective on the evolution of the high mountain cryosphere, thanks to accurate quantification of pluri-annual changes of volume and surface velocity. For instance, we could evidence : (1) a speed-up of the horizontal displacements since the 1990s in comparison with the previous decades; (2) an important surface lowering related to various melting processes (ice-core, thermokarst) for the three study sites; (3) relationships between the observed evolution and the contemporaneous climate warming, with a long-term evolution controlled by the warming of the ground and short-term changes that may relate to snow or precipitation or to the activity of the glacial-periglacial landforms.

How to cite: Cusicanqui, D., Rabatel, A., Bodin, X., Vincent, C., Thibert, E., Duvillard, P. A., and Revil, A.: Using historical aerial imagery to assess multidecadal kinematics and elevation changes. Application to mountain permafrost in the French Alps., EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-16371, https://doi.org/10.5194/egusphere-egu21-16371, 2021.

Archival imagery dating back to the mid-twentieth century holds information that pre-dates urban expansion and the worst impacts of climate change.  In this research, we examine deep learning colorisation methods applied to historical aerial images in Japan.  Specifically, we attempt to colorize monochrome images of river basins by applying the method of Neural Style Transfer (NST).    First, we created RGB orthomosaics (1m) for reaches of 3 Japanese rivers, the Kurobe, Ishikari, and Kinu rivers.  From the orthomosaics, we extract 60 thousand image tiles of `100 x100` pixels in order to train the CNN used in NST.  The Image tiles were classified into 6 classes: urban, river, forest, tree, grass, and paddy field.  Second, we use the VGG16 model pre-trained on ImageNet data in a transfer learning approach where we freeze a variable number of layers.  We fine-tuned the training epochs, learning rate, and frozen layers in VGG16 in order to derive the optimal CNN used in NST.  The fine tuning resulted in the F-measure accuracy of 0.961, 0.947, and 0.917 for the freeze layer in 7,11,15, respectively.  Third, we colorize monochrome aerial images by the NST with the retrained model weights.  Here used RGB images for 7 Japanese rivers and the corresponding grayscale versions to evaluate the present NST colorization performance.  The RMSE between the RGB and resultant colorized images showed the best performance with the model parameters of lower content layer (6), shallower freeze layer (7), and larger style/content weighting ratio (1.0 x10⁵).  The NST hyperparameter analysis indicated that the colorized images became rougher when the content layer selected deeper in the VGG model.  This is because the deeper the layer, the more features were extracted from the original image.  It was also confirmed that the Kurobe and Ishikari rivers indicated higher accuracy in colorisation.  It might come from the fact that the training dataset of the fine tuning was extracted from these river images.  Finally, we colorized historical monochrome images of Kurobe river with the best NST parameters, resulting in quality high enough compared with the RGB images.  The result indicated that the fine tuning of the NST model could achieve high performance to proceed further land cover classification in future research work.

How to cite: Ishii, R., Carbonneau, P., and Miyamoto, H.: Colorisation of archival aerial imagery using deep learning, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-11925, https://doi.org/10.5194/egusphere-egu21-11925, 2021.

The use of aerial photography in topography has started in the first decades of the 20^th century. Remote sensed data have become indispensable for cartographers and GIS staff when doing large-scale mapping: especially topographic, orienteering and thematic maps. The use of UAVs (unmanned aerial vehicles) for this purpose has also become widespread for some years. Various drones and sensors (RGB, multispectral and hyperspectral) with many specifications are used to capture and process the physical properties of an examined area. In parallel with the development of the hardware, new software solutions are emerging to visualize and analyse photogrammetric material: a large set of algorithms with different approaches are available for image processing.

Our study focuses on the large-scale topographic mapping of vegetation and land cover. Most traditional analogue and digital maps use these layers either for background or highlighted thematic purposes. We propose to use the theory of OBIA – Object-based Image Analysis to differentiate cover types. This method involves pixels to be grouped into larger polygon units based on either spectral or other variables (e.g. elevation, aspect, curvature in case of DEMs). The neighbours of initial seed points are examined whether they should be added to the region according to the similarity of their attributes. Using OBIA, different land cover types (trees, grass, soils, bare rock surfaces) can be distinguished either with supervised or unsupervised classification – depending on the purposes of the analyst. Our base data were high-resolution RGB and multispectral images (with 5 bands).

Following this methodology, not only elevation data (e.g. shaded relief or vector contour lines) can be derived from UAV imagery but vector land cover data are available for cartographers and GIS analysts. As the number of distinct land cover groups is free to choose, even quite complex thematic layers can be produced. These layers can serve as subjects of further analyses or for cartographic visualization.

BK is supported by Application Domain Specific Highly Reliable IT Solutions” project  has been implemented with the support provided from the National Research, Development and Innovation Fund of Hungary, financed under the Thematic Excellence Programme TKP2020-NKA-06 (National Challenges Subprogramme) funding scheme.

MP and FV are supported by EFOP-3.6.3-VEKOP-16-2017-00001: Talent Management in Autonomous Vehicle Control Technologies – The Project is financed by the Hungarian Government and co-financed by the European Social Fund.

How to cite: Kovács, B., Pál, M., and Vörös, F.: Object-based image segmentation in photogrammetry for cartographic use, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-11843, https://doi.org/10.5194/egusphere-egu21-11843, 2021.

Ephemeral rivers hydromorphological processes are intermittent and many times of fast response. Therefore they remain still quite unknown. The geomorphological mapping of river forms and geomorphological units is a useful tool to recognize the evolution, changes and the response of river adjustments of hydrological events.

A diachronic geomorphological mapping has been done in some ephemeral rivers located in Ebro basin, Segura basin and Calabrian ephemeral rivers. We are presenting the specific results of six reaches distributed by the Ebro basin (Tudela, Reajo, Alpartir, Cariñena, Valcodo, Sosa and Seco). The first historical aerial image is that of the American Flight B of 1956-57, another of the mid 80’s, the last official ortophotography available (around 2017), and a specific flight with an unmanned aerial vehicle (UAV) done during the winter of 2019. An altimetry correction has been performed on the first two images.

Different categories have been identified within the channel (active channel, principal channel and secondary channel), the sediment bars (vegetated, scant vegetated and non-active paleo-bars), the deposits coming from bank failures or tributaries, rocky areas, exhumed old sediment areas, consolidated or unconsolidated granular bed. The categories were mapped at different scales depending on the image quality (for example, from ≤ 1/300 scale of the UAV to ≤ 1/1,000 scale of the American flight).

This evolutionary cartography allows comparing the geomorphology of each river reach among different dates, considering the different resolution of the images and its limitations (i.e. previously, the results were unified to compare among them), and relating to the fluvial processes and changes on the river and basin.

This research was funded by ERDF/Spanish Ministry of Science, Innovation and Universities—State Research Agency (AEI) /Project CGL2017-84625-C2-1-R; State Program for Research, Development and Innovation Focused on the Challenges of Society. 

How to cite: Ibisate, A., Ollero, A., García, J. H., Ortiz Martínez de Lahidalga, J., Sáenz de Olazagoitia, A., Hermoso, Y., Conesa-García, C., and Gómez-Gutiérrez, A.: Geomorphological evolution of ephemeral rivers through historical and UAVs images, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-6012, https://doi.org/10.5194/egusphere-egu21-6012, 2021.

In high alpine geomorphological research, different technologies are increasingly used to describe and monitor mass movements such as rock glaciers. Among them, the combination of Unmanned Aerial Vehicle (UAV) systems and Structure from Motion (SfM) techniques is gaining continuous interest due to rapid technological developments. In this study, we test the capability of repeated UAV surveys to accurately survey rock glacier deformation in the Lac des Vaux area, Valais Alps. The studied landform is located on a typical anthropic alpine environment in the Swiss Alps, where ski facilities and alpine tracks are a commonplace. A DJI Phantom 4 RTK UAV was flown twice in September 2019 and September 2020 to cover an area of about 0.25 km² with nearly 1000 images each time. Differential corrections using a Virtual Reference Station (VRS) provided image geotags with centimetre-level accurate 3-D coordinates, thereby allowing dispensing with ground control. High-resolution orthomosaics and high-density point clouds are derived from the UAV-RTK surveys using a standard SfM processing workflow. The corresponding point clouds' accuracy was evaluated and adjusted based on stable terrain, reducing the 3-D alignment errors to a mean of 0.02 m. Elevation changes and surface kinematics in the rock glacier complex and its margins were quantified using point cloud operations and image correlation techniques. The results indicate that the landform has at least eight different lobes with mean velocities ranging between 0.1 and 1.3 m yr^-1. The high-resolution analysis also permitted identifying moving lobes without morphological expression and small thermokarst depressions on the ski slope structure that traverses the rock glacier's active zone. Without relying on ground control, our approach achieves horizontal and vertical accuracies nearly as good as monitoring techniques using more traditional differential GNSS devices.

How to cite: Vivero, S. and Lambiel, C.: Rock glacier deformation using an Unmanned Aerial Vehicle (UAV) with RTK GNSS capability, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-14844, https://doi.org/10.5194/egusphere-egu21-14844, 2021.

River systems are areas that respond most rapidly to regional and / or local tectonic movements, with global climate changes and general basement fluctuations that occurred during the Quaternary period. The most important geomorphological units where these events can be observed are terrace systems, which are the result of deposition and erosion processes. In fluvial geomorphology research on terrace systems, modern technological innovations are used as well as conventional field methods. Especially low-cost Unmanned Aerial Vehicles (UAV) and modern photogrammetry methods are preferred because they both provide detailed and precise identification of terraces and high resolution topography outputs in spatial and temporal terms.
This study aims to put a comprehensive mapping of the terrace systems observed in an area of 1.27 km² around Gemiciköy (Bilecik Province), which is located in the middle part of the Sakarya River valley, the largest river in Northwest Anatolia. Accordingly, we used the Structure from Motion (SfM) method which is based on photogrammetric principles and UAV. During the study, 582 images taken from a height of 100 m and having 80% overlap in line with the flight plans by using the DJI Mavic Mini UAV model were evaluated in Agisoft Metashape Professional software. With the use of image processing algorithms, the dense point cloud was first obtained, and then the orthomosaic and digital surface model with 3.29 cm resolution was produced. Two terrace levels (+10 m and +19 m) detected with digital surface modeling, and these were verified by stratigraphic and sedimentological observations made in the field.
As a result, low-cost UAV technologies are quite useful in terms of providing more detailed monitoring, mapping and analysis of river environments, together with the production of sensitive and high resolution topography data required in modern fluvial geomorphology research.

How to cite: Karakoca, E. and Uncu, L.: Mapping of Quaternary river terrace landforms with Unmanned Aerial Vehicle (UAV): A case study in the Sakarya River, NW Turkey, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-7166, https://doi.org/10.5194/egusphere-egu21-7166, 2021.