Geoscience Applications Research Articles

Progress and perspectives of point cloud intelligence

ABSTRACT With the rapid development of reality capture methods, such as laser scanning and oblique photogrammetry, point cloud data have become the third most important data source, after vector maps and imagery. Point cloud data also play an increasingly important role in scientific research and engineering in the fields of Earth science, spatial cognition, and smart cities. However, how to acquire high-quality three-dimensional (3D) geospatial information from point clouds has become a scientific frontier, for which there is an urgent demand in the fields of surveying and mapping, as well as geoscience applications. To address the challenges mentioned above, point cloud intelligence came into being. This paper summarizes the state-of-the-art of point cloud intelligence, with regard to acquisition equipment, intelligent processing, scientific research, and engineering applications. For this purpose, we refer to a recent project on the hybrid georeferencing of images and LiDAR data for high-quality point cloud collection, as well as a current benchmark for the semantic segmentation of high-resolution 3D point clouds. These projects were conducted at the Institute for Photogrammetry, the University of Stuttgart, which was initially headed by the late Prof. Ackermann. Finally, the development prospects of point cloud intelligence are summarized.

Generating Gridded Gross Domestic Product Data for China Using Geographically Weighted Ensemble Learning

Gridded gross domestic product (GDP) data are a crucial land surface parameter for many geoscience applications. Recently, machine learning approaches have become powerful tools in generating gridded GDP data. However, most machine learning approaches for gridded GDP estimation seldom consider the geographical properties of input variables. Therefore, in this study, a geographically weighted stacking ensemble learning approach was developed to generate gridded GDP data. Three algorithms—random forest, XGBoost, and LightGBM—were used as base models, and the linear regression in stacking ensemble learning was replaced by geographically weighted regression to locally fuse the three predictions. A case study was conducted in China to demonstrate the effectiveness of the proposed approach. The results showed that the proposed GDP downscaling approach outperformed the three base models and traditional stacking ensemble learning. Meanwhile, it had good predictive power on county-level GDP test data with R2 of 0.894, 0.976, and 0.976 for the primary, secondary, and tertiary sectors, respectively. Moreover, the predicted 1 km gridded GDP data had a high accuracy (R2 = 0.787) when evaluated by town-level GDP data. Hence, the proposed GDP downscaling approach provides a valuable option for generating gridded GDP data. The generated 1 km gridded GDP data of China from 2020 are of great significance for other applications.

Towards automatic and accurate core-log processing

The analysis of rocks plays an important role in geological and petroleum-engineering problems. In these tasks, core (i.e., core-log) is a crucial element since it provides underlying information on the geophysical properties of the area. Thereby, works often leverage core data to assign the correct category from well-logs. Unfortunately, processing core is laborious and time-consuming; hence, its analysis takes long periods. For example, for a single well, a human could have to adjust thousands of core data. Besides, due to its nature, log analysis must handle noise and missing data. Given these issues, our goal is to propose an automatic (e.g., without any human intervention) and accurate core processing using the gamma-ray. To achieve this goal and additionally demonstrate the most promising pattern recognition strategies, we assess the effectiveness of several models to diagnostic the core-log. Such models include simple regressions, ensembles, gradient boosting, recurrence models and the relatively recent Transformer network. Our comprehensive evaluation is different from existing works on geoscience tasks, which evaluate a small number of models or variations of a single model. Particularly, we evaluate more than 200 unique models (i.e., models with different hyperparameters) and observe that deep learning techniques outperform other techniques by a large margin. Furthermore, we compare the compromise between predictive ability and computational cost of several models – a reliable information for real-time lithology, which is often overlooked by previous works. According to our results, we can effectively replace the manual (i.e., by a human expert – geologist) diagnostic of core by pattern recognition methods, mainly by the Transformer model, as it aligns the cores in accordance (i.e., in the same direction) with the geologist. More specifically, given the core and gamma-ray as input, the Transformer model outputs an adjusted core obtaining an R2 of 93,63, which indicates a fine-grained adjustment. We empirically demonstrate that the success behind Transform is its effectiveness in expressing large sequences of core and well-log. On the other hand, other models such as RNN, LSTM and GRU meet collapse when processing large sequences of core-log. We further confirm that Transformers are among the top-performance models and often surpass recurrence models on several geoscience applications such as lithology and log-shape classification, and prediction of oil production. Regarding the first task, the models successfully classify different facies categories such as coarse sandstone, medium sandstone, fine sandstone, siltstone, dolomite, limestone and mudstone. In these tasks, Transformer outperforms the widely-employed LSTM by up to 14 percentage points. Our empirical observations encourage the exploration of multiple models and suggest Transformers as strong baselines for future research on geoscience tasks. In this direction, we release all data and trained models used throughout the work. To the best of our knowledge, we are the first study exploring Transformer models on geoscience applications.

Efficiently simulating Lagrangian particles in large-scale ocean flows — Data structures and their impact on geophysical applications

Studying oceanography by using Lagrangian simulations has been adopted for a range of scenarios, such as the determining the fate of microplastics in the ocean, simulating the origin locations of microplankton used for palaeoceanographic reconstructions, and for studying the impact of fish aggregation devices on the migration behaviour of tuna. These simulations are complex and represent a considerable runtime effort to obtain trajectory results, which is the prime motivation for enhancing the performance of Lagrangian particle simulators. This paper assesses established performance enhancing techniques from Eulerian simulators in light of computational conditions and demands of Lagrangian simulators. A performance enhancement strategy specifically targeting physics-based Lagrangian particle simulations is outlined to address the performance gaps, and techniques for closing the performance gap are presented and implemented. Realistic experiments are derived from three specific oceanographic application scenarios, and the suggested performance-enhancing techniques are benchmarked in detail, so to allow for a good attribution of speed-up measurements to individual techniques. The impacts and insights of the performance enhancement strategy are further discussed for Lagrangian simulations in other geoscience applications. The experiments show that I/O-enhancing techniques, such as dynamic loading and buffering, lead to considerable speed-up on-par with an idealised parallelisation of the process over 20 nodes. Conversely, while the cache-efficient structure-of-arrays collection yields a visible speed-up, other alternative data structures fail in fulfilling the theoretically-expected performance increase. This insight demonstrates the importance of good data alignment in memory and caches for Lagrangian physics simulations.

Spherical radial basis function approximation of some physical quantities measured

The article treats the spherical interpolation and approximation, i.e., a way of measured data processing in case when the scalar data sampling is performed at nodes on the unit 2D sphere surface in the 3D Euclidean space (in general, on the (d−1)-dimensional sphere surface in the d-dimensional space). We use the spherical radial basis function interpolation with an inverse multiquadric basis function and a second degree polynomial in Cartesian coordinates as a trend.The formulae of this type can be useful in the interpretation of various physical measurements and have wide applications in geosciences, e.g., in the treatment of anisotropy of magnetic susceptibility data measured. We present the advantages of the formula chosen on numerical examples. However, in practical computation with high number of sampling nodes, the matrix of the system for determining interpolation coefficients may be ill-conditioned. Then the standard double precision LU factorization does not give a reliable solution due to the effect of round-off error accumulation and some more sophisticated means of solving the system are to be used.

The impacts of reducing atmospheric and oceanic de-aliasing model error on temporal gravity field model determination

SUMMARYThe Gravity Recovery and Climate Experiment (GRACE) and its successor GRACE Follow-On mission have been estimating temporal gravity field models for various geoscience applications since 2002. While the errors in atmospheric and oceanic de-aliasing (AOD) model are still the key limitation for GRACE-type missions especially taking advantages of improved laser ranging interferometer (LRI), this paper explores the realistic assumption of continuously reducing AOD model error in global scale and regional scale (e.g. ocean, Greenland, Qinghai-Tibet plateau and South America) in the anticipated future. For this, using a realistic orbit scenario and error assumptions both for instrument and background model errors, the 5-yr full-scale simulations are implemented in the context of 26 scenarios with different AOD model errors. Our overall simulation results demonstrate that: (1) when the AOD model errors are reduced in the global scale, the corresponding gain in temporal gravity field determination is significant, with the cumulative geoid height error decrease from 11.28 to 6.25 mm in spectral domain, as well as 26.3–65.2 per cent noise reduction in terms of mean RMS residuals over ocean in spatial domain. (2) When the AOD model errors are reduced in the regional scale, the improvements of temporal gravity field estimations are diverse in different simulation scenarios. The most notable noise reduction, with 22.9–43.9 per cent noise reduction in spatial domain, is observed when the AOD model error decreases in ocean. The limited noise reductions (0.1 per cent and 1.3 per cent respectively in spectral and spatial domain) are observed in the temporal gravity field estimations if the AOD model is refined in other typical regions such as the Qinghai-Tibet plateau, the Greenland and the South America. (3) However, when the satellites passing over the AOD model refined regions, the along-orbit range rate analysis indicates that, there are visible differences by about 50.0 nm s−1 in terms of range rate residuals as well as 11.0–48.5 nm s−1 in terms of the mean RMS of range rate residuals. These results reflect the benefits of reducing AOD model error in both global and regional scale for improving GRACE-type temporal gravity estimation, especially considering the development of LRI technology.

Reconstruction of Sparse Stream Flow and Concentration Time‐Series Through Compressed Sensing

AbstractMonitoring water quality at high frequency is challenging and costly. Compressed sensing (CS) offers an approach to reconstruct high‐frequency water quality data from limited measurements, given that water quality signals are commonly “sparse” in the frequency domain. In this study, we investigated the sparsity of stream flow and concentration time‐series and tested reconstruction with CS. All stream signals were sparse using 15‐min discrete time‐series transformed to the Fourier domain. Stream temperature, conductance, dissolved oxygen, and nitrate plus nitrite (NOx‐N) concentration were sparser than discharge, turbidity, and total phosphorus (TP) concentration. CS effectively reconstructed these signals with only 5%–10% of measurements needed. Stream NOx‐N and TP loads were well estimated with errors of −6.6% ± 3.8% and −9.0% ± 2.9% with effective sampling frequencies of 10 and 0.4 days, respectively. For broader applications in environmental geosciences and engineering domains, CS can be integrated with dimensionality reduction and optimization techniques for more efficient sampling schemes.

Deep prior in variational assimilation to estimate an ocean circulation without explicit regularization

Abstract Many applications in geosciences require solving inverse problems to estimate the state of a physical system. Data assimilation provides a strong framework to do so when the system is partially observed and its underlying dynamics are known to some extent. In the variational flavor, it can be seen as an optimal control problem where initial conditions are the control parameters. Such problems are often ill-posed, regularization may be needed using explicit prior knowledge to enforce a satisfying solution. In this work, we propose to use a deep prior, a neural architecture that generates potential solutions and acts as implicit regularization. The architecture is trained in a fully-unsupervised manner using the variational data assimilation cost so that gradients are backpropagated through the dynamical model and then through the neural network. To demonstrate its use, we set a twin experiment using a shallow-water toy model, where we test various variational assimilation algorithms on an ocean-like circulation estimation.

Carefully Choose the Baseline: Lessons Learned from Applying XAI Attribution Methods for Regression Tasks in Geoscience

Abstract Methods of explainable artificial intelligence (XAI) are used in geoscientific applications to gain insights into the decision-making strategy of neural networks (NNs), highlighting which features in the input contribute the most to a NN prediction. Here, we discuss our “lesson learned” that the task of attributing a prediction to the input does not have a single solution. Instead, the attribution results depend greatly on the considered baseline that the XAI method utilizes—a fact that has been overlooked in the geoscientific literature. The baseline is a reference point to which the prediction is compared so that the prediction can be understood. This baseline can be chosen by the user or is set by construction in the method’s algorithm—often without the user being aware of that choice. We highlight that different baselines can lead to different insights for different science questions and, thus, should be chosen accordingly. To illustrate the impact of the baseline, we use a large ensemble of historical and future climate simulations forced with the shared socioeconomic pathway 3-7.0 (SSP3-7.0) scenario and train a fully connected NN to predict the ensemble- and global-mean temperature (i.e., the forced global warming signal) given an annual temperature map from an individual ensemble member. We then use various XAI methods and different baselines to attribute the network predictions to the input. We show that attributions differ substantially when considering different baselines, because they correspond to answering different science questions. We conclude by discussing important implications and considerations about the use of baselines in XAI research. Significance Statement In recent years, methods of explainable artificial intelligence (XAI) have found great application in geoscientific applications, because they can be used to attribute the predictions of neural networks (NNs) to the input and interpret them physically. Here, we highlight that the attributions—and the physical interpretation—depend greatly on the choice of the baseline—a fact that has been overlooked in the geoscientific literature. We illustrate this dependence for a specific climate task, in which a NN is trained to predict the ensemble- and global-mean temperature (i.e., the forced global warming signal) given an annual temperature map from an individual ensemble member. We show that attributions differ substantially when considering different baselines, because they correspond to answering different science questions.

Corrections to: “Accurate computation of gravitational field of a tesseroid” by Fukushima (2018) in J. Geod. 92(12):1371–1386

An accurate method with conditional split, double exponential quadrature rule, and numerical differentiation has been proposed in the paper “Accurate computation of gravitational field of a tesseroid” (Fukushima in J Geod 92(12):1371–1386, https://doi.org/10.1007/s00190-018-1126-2, 2018) to compute the gravitational field (i.e. gravitational potential, gravitational acceleration vector, and gravity gradient tensor) of a tesseroid. This study presents the corrections for some formulas in the main paper and electronic supplementary material of Fukushima (J Geod 92(12):1371–1386, https://doi.org/10.1007/s00190-018-1126-2, 2018). Moreover, the FORTRAN subroutines gtess (or qgtess) and ggtess (or qggtess) in the original codes xtess.txt (or xqtess.txt) in double (or quadrature) precision provided by Fukushima (J Geod 92(12):1371–1386, https://doi.org/10.1007/s00190-018-1126-2, 2018) are revised. The revised parts have impacts on the calculation of these components of the gravitational acceleration vector (g_{varPhi } and g_{varLambda }) and gravity gradient tensor (varGamma _{varPhi varPhi }, varGamma _{varPhi varLambda }, varGamma _{varPhi H}, varGamma _{varLambda varLambda }, varGamma _{varLambda H}, and varGamma _{H H}). The revised FORTRAN codes xtess.f90 and xqtess.f90 in double and quadrature precision are presented at the GitHub website https://github.com/xiaoledeng/xtess-xqtess. These revised FORTRAN codes can accurately compute the gravitational field of a tesseroid in double and quadrature precision no matter the computation point is located outside, near the surface of, on the surface of, or inside the tesseroid. They can be applied to calculate the gravitational field of the different layers (e.g. atmosphere, topography, crust, and mantle) of the Earth or other celestial bodies, which helps investigate the various geoscience applications, e.g. geoid determination in geodesy and gravity interpretation in geophysics.

New topic horizons for drone systems and applications

Summary This journal (Drone Systems and Applications; DSA) conducted a targeted “horizon scan” during 2022 within our team of editors and associate editors. We asked— Which research areas currently under-represented in Drone Systems and Applications would you like to see more heavily represented in the future? The process highlighted five areas of interest and potential growth: Drones in the geosciences Aquatic drones Ground drones Drones within calibration/validation experiments Drones and computer vision Over the past two years (2020–22), the journal has published over 50 papers with a strong leaning towards aerial drones for ecology and also with an engineering focus. DSA is keen to receive new submissions addressing the five highlighted areas, which lie firmly within the aims and scope of the journal. Further to the horizon scan, we propose two special collections for the coming year—one addressing drone applications ( drones in geoscience applications) and a second addressing drone systems ( aquatic drone systems). We would like to hear from scientists and practitioners in these fields as both contributors and (or) collection editors.

Analysis of the precision of different digital models of global and local elevations in continental Ecuador

Currently, DEMs are inputs in multiple geoscience applications. Ecuador presents three regions with topographic characteristics that influence the performance of the DEM on altimetric precision. The main objective has been to analyze the precision of five global models and the official local model of the country, using GPS points to determine the level of adjustment of the DEM in the regions of Ecuador. The cities of Macas (Amazonian basin), Quito (Highlands) and Guayaquil (Coastal lowlands) were considered as study areas for each region. Statistics and spatial correlation / dispersion graphs were obtained with the R software, also with QGIS longitudinal profiles of the models were performed. According to the statistics calculated, the local SIG-TIERRAS model presents more auspicious results in the Highlands and Coast of the country, with an RMSE of 2,498, 1,556 meters, while in the Amazon the ALOS model is slightly better than the local DEM, with 5,792 meters of RMSE.

A single-molecule study on polymer fluid dynamics in porous media.

Understanding the dynamic behavior of polymeric fluids in porous media is essential for vast geoscience applications, particularly enhanced oil recovery and polymer-enhanced soil washing, to clean up soil contamination. During the past decades, the behavior of polymeric fluids in microscopic space has only been investigated using ensemble-averaged experimental methods in which a bulk phase behavior of the fluids characterizes flow mechanisms. Multiple flow mechanisms have been proposed based on ensemble-averaged data; however, microscale characterization of the interactions between polymers and solid surfaces and the mechanisms governing polymer retention and permeability reduction as well as the reversibility of polymer retention are lacking, resulting in a limited understanding of the flow mechanisms. Here we report direct visualization and multi-scale characterization of the dynamic behavior of polymer molecules in a representative porous medium by integrating microfluidics with single-molecule imaging. We demonstrate that the polymers' adsorption, entrapment and hydrodynamic retention contribute to their overall retention in porous media. Our study illustrates how microfluidics can help in understanding the dynamic behavior of polymers, their interactions with the solid/fluid interface and their effects on flow properties. Additionally, it demonstrates the role of microfluidic platforms in providing a more representative and accurate model for polymer retention and permeability reduction in porous media. The obtained insights encourage the development of improved models that better capture the behavior of complex fluids in confined environments and have significant implications for a wide range of applications in geoscience, materials science, and rheology.

Generating Sentinel-2 all-band 10-m data by sharpening 20/60-m bands: A hierarchical fusion network

Earth observations from the Sentinel-2 mission have been extensively accepted in a variety of land services. The thirteen spectral bands of Sentinel-2, however, are collected at three spatial resolutions of 10/20/60 m, and such a difference brings difficulties to analyze multispectral imagery at a uniform resolution. To address this problem, we developed a hierarchical fusion network (HFN) to sharpen 20/60-m bands and generate Sentinel-2 all-band 10-m data. The deep learning architecture is used to learn the complex mapping between multi-resolution input and output data. Given the deficiency of previous studies in which the spatial information is inferred only from the fine-resolution bands, the proposed hierarchical fusion framework simultaneously leverages the self-similarity information from coarse-resolution bands and the spatial structure information from fine-resolution bands, to enhance the sharpening performance. Technically, the coarse-resolution bands are super-resolved by exploiting the information from themselves and then sharpened by fusing with the fine-resolution bands. Both 20-m and 60-m bands can be sharpened via the developed approach. Experimental results regarding visual comparison and quantitative assessment demonstrate that HFN outperforms the other benchmarking models, including pan-sharpening-based, model-based, geostatistical-based, and other deep-learning-based approaches, showing remarkable performance in reproducing explicit spatial details and maintaining original spectral features. Moreover, the developed model works more effectively than the other models over the heterogeneous landscape, which is usually considered a challenging application scenario. To sum up, the fusion model can sharpen Sentinel-2 20/60-m bands, and the created all-band 10-m data allows image analysis and geoscience applications to be authentically carried out at the 10-m resolution.

CAFE: A Cross-Attention Based Adaptive Weighting Fusion Network for MODIS and Landsat Spatiotemporal Fusion

Dense medium-resolution images play an important role in time-series geoscience applications. However, due to technical limitations, remote sensing imaging systems inevitably trade off temporal frequency and spatial swaths, resulting in difficulties to acquire images simultaneously with high spatial and temporal resolution. To overcome this limitation, under the framework of residual learning, we propose a Cross-attention based Adaptive weighting Fusion nEtwork (CAFE) for MODIS-Landsat spatiotemporal fusion to generate dense medium-resolution images. Based on the cross-attention mechanism, we propose multi-channel separated cross-attention and full-feature joint cross-attention blocks to enhance spatial resolution and retain spectral signatures from the perspectives of band-wise processing and full-feature joint processing, respectively. The adaptive temporal difference weighting mechanism is proposed to improve the ability to capture dynamic land surface changes. Besides, we employ an adaptive fusion loss function to constrain the network training. Experimental results indicate that the developed method is superior to several existing algorithms in terms of visual evaluation and quantitative evaluation and it can generate high-quality fusion results in scenarios of both subtle and dramatic temporal changes. Codes will be available at https://github.com/LiupengLin/CAFE.

Editorial of Special Issue “Advances and Applications in Computational Geosciences”

In geoscientific fields, mathematical modelling, numerical analysis, visualization, simulation, and other computational techniques have become necessary to improve our understanding of phenomena and evolution of the Earth [...]

Local Embedded Discrete Fracture Model (LEDFM)

The study of flow in fractured porous media is a key ingredient for many geoscience applications, such as reservoir management and geothermal energy production. Modelling and simulation of these highly heterogeneous and geometrically complex systems require the adoption of non-standard numerical schemes. The Embedded Discrete Fracture Model (EDFM) is a simple and effective way to account for fractures with coarse and regular grids, but it suffers from some limitations: it assumes a linear pressure distribution around fractures, which holds true only far from the tips and fracture intersections, and it can be employed for highly permeable fractures only. In this paper we propose an improvement of EDFM which aims at overcoming these limitations computing an improved coupling between fractures and the surrounding porous medium by (a) relaxing the linear pressure distribution assumption, (b) accounting for impermeable fractures modifying near-fracture transmissibilities. These results are achieved by solving different types of local problems with a fine conforming grid, and computing new transmissibilities (for connections between fractures and the surrounding porous medium and those through the porous medium itself near to the fractures). Such local problems are inspired from numerical upscaling techniques present in the literature. The new method is called Local Embedded Discrete Fracture Model (LEDFM) and the results obtained from several numerical tests confirm the aforementioned improvements.

THE COMPARISON OF DIFFERENT LIDAR ACQUISITION SOFTWARE ON IPAD PRO M1 2021

Abstract. The application of Apple LiDAR is now being researched in a variety of fields. Several research projects have been conducted to provide an answer, including those focused on heritage documentation, forest inventory, and geoscience applications. In cave, it is common practice to photograph with a digital camera equipped with photogrammetric techniques and a terrestrial laser scanner. The focus of this study is on how the iPad Pro M1 2021 can generate a point cloud that describes the graffiti on the cave walls. This investigation was carried out in Barong Cave, West Java, Indonesia. Scanning with iPad Pro M1 2021 was completed using three applications: 3D Scanner App, EveryPoint, and SiteScape. Point cloud of each application is analyzed using M3C2 plugin in CloudCompare. Based on this research, SiteScape is the best application to use. It can be concluded that the LiDAR iPad is suitable for use in caves and can capture detailed information about cave walls.

Quantifying the Permeability Enhancement from Blast-Induced Microfractures in Porphyry Rocks Using a Cumulant Lattice Boltzmann Method

The permeability of rocks is important in a range of geoscientific applications, including hbox {CO}_{2} sequestration, geothermal energy extraction, and in situ mineral recovery. This work presents an investigation of the change in permeability in porphyry rock samples due to blast-induced fracturing. Two samples were analysed before and after exposure to stress waves induced by the detonation of an explosive charge. Micro-computed tomography was used to image the interior of the samples at a pixel resolution of 10.3,mu m. The images were segmented into void, matrix, and grain to help quantify the differences in the rock samples. Following this, they were binarised as void or solid and the cumulant lattice Boltzmann method (LBM) was applied to simulate the flow of fluid through the connected void space. A correction required with the use of inlet and outlet reservoirs in computational permeability assessment was also proposed. Interrogation of the steady-state flow field allowed the pre- and post-loading permeability to be extracted. Conclusions were then drawn as to the effectiveness of blasting for enhancing fluid accessibility via the generation of microfractures in the rock matrix within the vicinity of a detonated charge. This paper makes contributions in three fundamental areas relating to the numerical assessment of permeability and the enhancement of fluid accessibility in low-porosity rocks. Firstly, a correction factor was proposed to account for the reservoirs commonly imposed on digitised rock samples when investigating sample permeability through numerical methods. Secondly, it validates the benefits of the LBM in handling complex geometries that would be intractable with conventional computational fluid dynamics methods that require body-fitted meshing. This is done with a novel implementation of the cumulant LBM in the open-source TCLB code. Finally, the improvement in fluid accessibility in low-permeability rock samples was shown through the assessment of multiple regions within two blasted samples. It was found that the blast-induced loading can generate extended microfractures that results in multiple orders of magnitude of permeability enhancement if the target rock possesses existing weaknesses and/or mineralisation.

Machine-Learning Applications in Geosciences: Comparison of Different Algorithms and Vegetation Classes’ Importance Ranking in Wildfire Susceptibility

Susceptibility mapping represents a modern tool to support forest protection plans and to address fuel management. With the present work, we continue with a research framework developed in a pioneristic study at the local scale for Liguria (Italy) and recently adapted to the national scale. In these previous works, a random-forest-based modeling workflow was developed to assess susceptibility to wildfires under the influence of a number of environmental predictors. The main novelties and contributions of the present study are: (i) we compared models based on random forest, multi-layer perceptron, and support vector machine, to estimate their prediction capabilities; (ii) we used a more accurate vegetation map as predictor, allowing us to evaluate the impacts of different types of local and neighboring vegetation on wildfires’ occurrence; (iii) we improved the selection of the testing dataset, in order to take into account the temporal variability of the burning seasons. Wildfire susceptibility maps were finally created based on the output probabilistic predicted values from the three machine-learning algorithms. As revealed with random forest, vegetation is so far the most important predictor variable; the marginal effect of each type of vegetation was then evaluated and discussed.