A Review of Digital Elevation Model Super Resolution

The scarcity of high-resolution urban digital elevation model (DEM) datasets, particularly in certain developing countries, has posed a challenge for many water-related applications such as flood risk management. A solution to address this is to develop effective approaches to reconstruct high-resolution DEMs from their low-resolution equivalents that are more widely available. However, the current high-resolution DEM reconstruction approaches mainly focus on natural topography. Few attempts have been made for urban topography, which is typically an integration of complex artificial and natural features. This study proposed a novel multi-scale mapping approach based on convolutional neural network (CNN) to deal with the complex features of urban topography and to reconstruct high-resolution urban DEMs. The proposed multi-scale CNN model was firstly trained using urban DEMs that contained topographic features at different resolutions, and then used to reconstruct the urban DEM at a specified (high) resolution from a low-resolution equivalent. A two-level accuracy assessment approach was also designed to evaluate the performance of the proposed urban DEM reconstruction method, in terms of numerical accuracy and morphological accuracy. The proposed DEM reconstruction approach was applied to a 121 km2 urbanized area in London, United Kingdom. Compared with other commonly used methods, the current CNN-based approach produced superior results, providing a cost-effective innovative method to acquire high-resolution DEMs in other data-scarce regions.

High-resolution Digital Elevation Model (DEM) data provides essential information for pluvial flood simulation. Although the increased accessibility and quality of publicly available DEM datasets can facilitate geospatial analysis at various scales, existing DEM datasets with global coverage mostly lack sufficient spatial resolution for pluvial flood simulations, which require detailed topographic information to be included in the simulation. Simulating flood scenarios with low-resolution DEMs (>30m) can result in substantial deviations from real cases. This issue becomes even more severe for flood-prone areas in data-scarce developing countries.Image super-resolution is a technique for reconstructing low-resolution information into high-resolution data. Various deep-learning models have been employed for this task, primarily focusing on generating high-resolution natural-colour images. However, the effects of these deep learning models on enhancing the resolution of DEM data have not been extensively investigated. One of the state-of-the-art super-resolution models, the Residual Channel Attention Network (RCAN), has gained popularity due to its accuracy and efficiency. Leveraging publicly available low-resolution global DEM data and high-resolution regional DEM data, this study assesses the performance of RCAN models in a DEM super-resolution task. The experimental results suggest that, compared to conventional interpolation methods, the tested RCAN model exhibits superior performance in constructing high-resolution DEM data. The generated super-resolution DEM data were then tested in pluvial flood simulations and achieved substantially higher realism in modelling floodwater distribution. The proposed method for constructing super-resolution DEMs opens up the possibility of simulating flooding at hyper-resolution globally.

The spatial variability of input parameters plays a crucial role in the interpretation of geomorphic indices, with digital elevation models (DEMs) being the primary data source. However, the influence of DEM resolution on these indices has rarely been investigated. This study investigated the influence of DEM resolution on the assessment of tectonic activity using the normalized stream length–gradient (SLk) index, which reflects variations along river profiles. The SLk index is sensitive to changes in river gradients that may result from active faulting or differential uplift, making it a valuable tool for identifying zones of active tectonic deformation. Therefore, understanding the impact of DEM resolution on SLk analysis is critical for accurately detecting and interpreting subtle tectonic signals, particularly in intraplate regions where deformation is slow and geomorphic expressions are faint and discontinuous. By comparing high-resolution LiDAR-derived DEMs (L-DEMs) and low-resolution topographic map-derived DEMs (T-DEMs), we analyzed the SLk index distributions along the Yangsan Fault, Korean Peninsula, an intraplate setting with Quaternary activity. According to the results, SLk anomalies derived from L-DEMs had a continuous distribution along the fault, closely aligning with known surface ruptures and indicating active tectonic deformation. In contrast, SLk anomalies derived from T-DEMs were sporadic and less continuous, especially in low-relief landscapes such as alluvial fans and floodplains, highlighting the limitations of T-DEMs in detecting fault-related features. High-resolution DEMs were better able to capture finer-scale geomorphic features, such as fault scarps, deflected streams, and lineaments associated with active tectonics, providing a more comprehensive view of fault-related deformation. This discrepancy highlights the importance of resolution choice in tectonic assessments, as low-resolution DEMs may underestimate the tectonic activities of intraplate faults by missing subtle topographic variations. While the choice of DEM resolution may depend on study area, scope, and data availability, high-resolution DEMs are critical for identifying tectonic activity in intraplate regions where geomorphic features of faulting due to slow deformation are subtle and dispersed.

Generating high-resolution DEMs in mountainous regions using ICESat-2/ATLAS photons

Flood models predict inundation extents, and can be an important source of information for flood risk studies. Accurate flood models require high resolution and high accuracy digital elevation models (DEM); current global DEMs do not capture the topographic details in floodplains, and this often leads to inaccurate prediction of flood extents by flood models. Flood extents obtained from remotely sensed data provide indirect information about topography. Here, we attempt to use this information along with model predictions to produce better floodplain topography. The algorithm we describe is a two-step process: first, we reduce the noise along the observed flood boundaries for all particles. Then, the model predictions from these modified DEMs are assimilated with observations using a particle batch smoother. We implemented the algorithm for a synthetic test case. For the nominal case, we observed a significant improvement in accuracy in terms of RMSE (35% reduction), bias (20%) and standard deviation (40%). We conducted sensitivity analysis by using priors of varying bias (0.5 m, 1 m, 2 m) and standard deviation (1 m, 2 m, 4 m). The bias reduced to ~0.5 m or below in all the cases: the reduction in bias varied from 11% to 76%. The standard deviation of errors in the final estimate was almost half of the prior: the reduction varied from 40% to 49%. The reduction in RMSE ranged between 35% and 67%. For the case with 2 m bias and 4 m standard deviation (SRTM-like error levels), bias went down to 0.48 m (76% reduction), and standard deviation reduced to 2.24 m (44% reduction). Flood inundation maps produced from the final estimate DEMs also improved on its prior. For the 2 m bias cases, true positive rate (TPR) for peak inundation went from ~30% to more than 57% in all three cases. The algorithm produces promising results, and this type of analysis can be performed in data-poor floodplains where high resolution DEMs do not exist.

A rapid high-resolution multi-sensory urban flood mapping framework via DEM upscaling

Machine Learning (ML) is extensively used in diverse topic domains, including geographical information. Despite its limitations, using Digital Elevation Models (DEMs) is gradually being considered in various operational applications. This study explores the application of machine learning algorithms, Shuttle Radar Topography Mission (SRTM) data with different resolutions, and data collected by Unmanned Aerial Vehicle (UAV) technology to produce high-resolution DEMs. The proposed construction approach is based on eight algorithms Linear Regression, Decision Tree, Random Forest, Ridge, Lasso, SVR, K-Neighbors Regressor, and XGB Regressor to deal with the complex features of urban topography to reconstruct high-resolution urban DEMs. The proposed algorithms were applied to two different study areas in Egypt. The results were contrasted with a reference DEM obtained from the ground measurement data (UAV). The numerical accuracy and terrain feature preserving effects of the Linear Regression algorithm in the first study area can generate reconstructed DEMs that better match the Reference DEMs, show lower mean absolute error (MAE), mean square error (MSE), and improve the accuracy of the terrain parameters by the overall fitness R2 of .971. The results showed that the Linear Regression algorithm is the most accurate with an R2 of .985. Compared to other commonly used methods, the current proposed approach offers a cost-effective and innovative method for acquiring high-resolution DEMs in other data-scarce regions, resulting in superior results. Our research is a comprehensive examination of geographical artificial intelligence (Geo AI), which is a term that refers to the processing of geographic information.

A 10-metre meshed high resolution Digital Elevation Model (DEM) has been integrated with Geographic Information Systems (GIS). High resolution DEM plays a crucial role in future GIS applications such as constructing 3D systems in Japan, but it is not sufficient for precise requirements from various applications. Also, a one-metre high resolution pansharpen image from the IKONOS satellite has been integrated with a digital map on the GIS system at the same time. IKONOS images cover 3-dimensional surfaces constructed by 10 m DEM. A precise GIS map of Kitakyushu City in Japan has been used as a digital map, which is quite accurate.

Abstract. High-resolution (HR) digital elevation models (DEMs), such as those at resolutions of 1 and 3 meters, have increasingly become more widely available, along with lidar point cloud data. In a natural environment, a detailed surface water drainage network can be extracted from a HR DEM using flow-direction and flow-accumulation modeling. However, elevation details captured in HR DEMs, such as roads and overpasses, can form barriers that incorrectly alter flow accumulation models, and hinder the extraction of accurate surface water drainage networks. This study tests a deep learning approach to identify the intersections of roads and stream valleys, whereby valley channels can be burned through road embankments in a HR DEM for subsequent flow accumulation modeling, and proper natural drainage network extraction.

At the end of the nineties the emergence of high resolution (1 m) digital elevation models (DEMs) settled the context of high precision geomorphological analysis. These new elevation models permitted to reveal structures that remained heretofore undetectable. Earth scientists henceforth benefit from a field of analysis with a textural richness that was never attained before. However, the complexity and the volume of the data reveal a series of questions and problems. The storage size has increased, the computational processes have become heavier, and the visual or digital interpretation has become more complex. Moreover, these new models make it possible to characterize and analyse much smaller phenomena than previously. Traditional DEMs with resolutions ranging from 10 m to 90 m can be used to analyse a valley or a hillside. Transposed to a cartographic scale, this corresponds at best to a 1 : 25 000 ratio. As for high resolution DEMs, they show much more detailed structural levels and can be used to analyse geomorphological features of 2 – 3 meters, this corresponding to scales ranging from 1 : 10 000 to 1 : 1 000. Yet, in the abundance offered by this growing resolution, large geomorphological structures are still present, including the finer structures. They are even the actuators of processes relevant to larger cartographic scales. Consequently, high resolution DEMs contain a multitude of structures, which exist throughout their interactions with other structures at other scales. This is the context of the present study. Geomorphometry – the quantitative counterpart of exploratory geomorphology – permits to explore and quantify a wide range of shapes and terrain indicators. At higher resolution however, the methods of this discipline can hardly be used. Geomorphometrical methods are based on a geometric model (a quadratic surface) and few of these methods can be applied it in a multiscale context. Furthermore inappropriate techniques are frequently used, hence the idea to move to a multiscale approach called the wavelet transform. The latter had previously been explored by few researchers within the geomorphometry community, but never thoroughly to micro- and to mesoscales. Due to the non-stationarity of DEMs, the wavelet transform was preferred to the Fourier transform in order to decompose DEMs into multiscale spaces. This facilitates a coherent navigation from scale to scale, but also makes new scale specific phenomena emerge for different frequencies. The wavelet transform is a technique widely used in image analysis. It allows decomposing a signal according to its frequency components, but also according to the position of the frequencies in the signal. Its multi-scale capacity is an effective analytical tool in multiple domains. More particularly in geomorphology, structural components – specific to a specific phenomenon – are well determined in these sub-spaces specific to the scale continuum. Finally an in-depth analysis of the phenomena enabled us to understand processes and their and their phenomenological inter-dependencies. In order to understand the effects and outcomes of the approach we developed an artificial landslide. We then computed some profiles and analysed the autocorrelation, slope attenuation and local fractal indicator. The resulting high-pass information of the wavelet transform has also been analysed and filtered using several types of filters. In a case study we used a real-world landslide to validate the transform and to understand its impact on geological structures. Within this case-study we conducted a web-based survey that allowed the participants to analyse the landslide using wavelet results and to make comments on the potential of the wavelet transform in the field of geomorphometry. Moreover, important contributions of this thesis are new algorithms that allow the illustration of the structural coherence in relation to each subspace. These are based on the theory of vision of Marr and on structure tensors. The results of our studies show a high consistency. The wavelet transform thereby extends the range of tools in geomorphometry. The different structural scale levels show that such these methods are needed to better understand the phenomenology of geomorphological processes.

Gridding Mars Orbiter Laser Altimeter data with GMT: effects of pixel size and interpolation methods on DEM integrity

Digital elevation models (DEM) or digital terrain elevation models (DTED) are widely used in radio propagation simulation and prediction. The resolution of DEM's is getting higher and higher with the latest resolution of centimeters generated by LIDAR (light detection and ranging). These high resolution DEM's provide more realistic representation of the environments and will enhance the accuracy of propagation modeling results. But they also have some drawbacks. First, the storage of high resolution DEM's will cost large computer disk space and slowdown the I/O process. Second, not all propagation scenarios need high resolution DEM's depending on the frequency and range. Third, the DEM's do not explicitly have the three dimensional (3D) information about topographic structures, such as ridges, which is physically important in diffracted field calculations. For example, 3D ridges are key contributors of diffracted field in certain mountainous regions. But many widely used and DEM-based methods (e.g., knife-edge models) ignore the 3D features. Our previous work has shown that the orientation and interior angle of a ridge can cause 3dB difference in path loss prediction for a single ridge. For two ridges, the difference is almost doubled.

Quantifying landscape form can provide crucial insight into the interactions between tectonics and climate. River long profile morphology, quantified by metrics such as channel steepness, is the most commonly used tool to investigate topographic form, with many studies relating long profile morphology to uplift rate, precipitation, sediment properties, or lithology, for example. River long profiles record the signal of external forcing over large spatial scales (i.e. tens of kilometres). This has many advantages: for example, it is a convenient scale for analysing variations in large-scale processes, such gradients in tectonic uplift. It also means that high resolution digital elevation models (DEMs) are not required and therefore river long profiles can be extracted globally. However, analysis of river long profiles over tens of kilometres can also result in signal smoothing and subsequent loss of finer scale tectonic or climatic signatures encoded into the landscape.Tectonic and climatic processes do not only leave their fingerprint in the long profiles of rivers. Hilltops, hillslopes, and valleys make up the majority of Earth’s landscapes by area, yet their morphology has received much less attention than that of rivers. This is in part due to the difficulty in accurately extracting hilltops and valley morphology from DEMs, especially on a global scale. Here, I show that we can now extract hilltop and valley metrics from high-resolution (< 15 m) DEMs over orogenic to continental scales using new topographic analysis techniques and high-performance computing facilities. I argue that by combining hilltop, hillslope, and valley metrics, we can obtain more information about tectonic and climatic processes than from river profiles alone.

The bistatic acquisition mode of the TanDEM-X radar satellite mission provides a new reliable source for generating high-resolution (~10m) digital elevation models (DEMs) of active volcanoes. For high relief-mountainous areas with steep slopes - like stratovolcanoes - geometric distortions considerably affect the accuracy of the DEMs. The TanDEM-X system allow for the first time access to high-resolution DEMs for both orbits (ascending and descending) which, by using an adaptive approach to combine multiple acquisitions from different viewing angles, considerably increase the accuracy of the DEM over difficult terrain like active volcanoes. By integrating these DEMs into standard numerical flow modeling, we are able to enhance the performance of standard terrain-dependent volcanic flow models and reduce the uncertainty that inevitably arises in application of topographic models of varying resolutions. Our study pushes the limits on where and how SAR data can be used to derive accurate DEMs under extreme conditions.

Georegistration is the assignment of 3-D coordinates to the pixels of an image. It is necessary for aligning imagery with map data, fusing images from different sensors, and geolocating moving objects. Georegistration requires recovery of the exterior camera orientation or pose - that is, position and attitude - for every image in the motion imagery sequence, and often the recovery of the intrinsic camera parameters of focal length, sensor pixel aspect ratio, and radial distortion. We have developed a real- time, automated solution to this problem. It is based on the registration of actual images to predicted images from a high-resolution digital elevation model (DEM). Beginning from an initial camera estimate for the first image, the algorithm iterates on the process of forming a predicted image, registering it to the actual image, and then refining the camera estimate based on the registration results. The resulting camera model forms the initial camera estimate of the next image. In this way, the camera model of each image is recovered as well as the platform motion, and the resulting camera models are used with the DEM to stabilize and georegister the motion imagery. The algorithm works with a wide variety of DEMs, including high-resolution LIDAR and low-resolution USGS DEMs. Implemented on a dual quad-core PC, the software georegisters one gigapixel of imagery per second. Results are presented for wide area motion imagery, full motion video, and thermal infrared video, along with applications to surveillance and border security.