Excited to share our research #article entitled "Regional‑scale spatiotemporal landslide probability assessment through machine learning and potential applications for operational warning systems: a case study in Kvam (Norway)"! Our study proposes a method that uses a #MachineLearning approach for both spatial and temporal #landslide forecasting at the regional scale. The aim is to overcome the traditional static applications of machine learning and demonstrate the applicability of a novel model aimed at spatiotemporal landslide #hazard mapping, with perspectives of applications to early warning systems. Check it out in the Landslide journal! Open Access: https://lnkd.in/dF_KN7kj Thanks to the team: Ascanio Rosi, Samuele Segoni, Luca Piciullo, Zhongqiang Liu, Riccardo Fanti. Machine Intelligence and Slope Stability Laboratory NGI - Norwegian Geotechnical Institute Università degli Studi di Firenze #artificialintelligence #RandomForest #openaccess #slopestability #earlywarning
Nicola Nocentini’s Post
More Relevant Posts
-
I'm very excited about the results we are achieving through Nimbus Insights, applying advanced data science and remote sensing techniques to geospatial data. In this analysis, we combined the results of a principal component analysis applied to surface reflectance data from the Harmonized Sentinel-2 MSI collection with elevation data from Alos Palsar (Hi-Res Terrain Corrected). Through meticulous analysis of the eigenvector matrix, factor loadings, and the images produced by the newly identified factors, we have developed color compositions that elucidate various characteristics of the study area. This analysis enables the extraction of structural, mineralogical, textural, and geomorphological information, proving to be very useful for structural mapping, geological studies, and mineral exploration, especially in areas with sparse vegetation. The analyzed area is located in Iran, in the Hormozgan province, located in the south of Iran, known for its coasts and islands in the Persian Gulf, as well as its historical and cultural significance. The chosen area encompasses the ancient Amir mine and the chromite occurrences of Kuh-E-Sorkh. I would like to extend my gratitude to my colleagues at Nimbus Insights, Murillo Costa and Ramon Arouca Jr., for their collaboration in this significant endeavor of applying technological advancements to geosciences. Interactive 3D models of our findings are hosted in the cloud and can be accessed through the links provided below. For demonstration, we have selected three color compositions and one RGB image. RGB 3D Model Link: https://p3d.in/BxHOr Color Composite 3D Model 1: https://p3d.in/xuJNT Color Composite 3D Model 2: https://p3d.in/3SR5n Color Composite 3D Model 3: https://p3d.in/Br29S
To view or add a comment, sign in
-
Professor and Head of Data Science for the Environment and Sustainability, Digital Environments Research Institute (DERI), Queen Mary University of London, and Carbonate Expert at Applied Stratigraphix
I invite you to check our work on using deep learning for satellite data denoising on Mars: this is my PhD student Robert Platt ‘s first paper! #mars #deeplearning #machinelearning #geosciences
Very excited to share that my first-author paper "Noise2Noise Denoising of CRISM Hyperpsectral data" has been accepted to the ICLR Machine Learning for Remote Sensing (ML4RS) workshop! CRISM data has revolutionised our understanding of Martian surface mineralogy, but degrading quality over time has limited its use. Our method (N2N4M) rapidly denoises CRISM imagery to allow for more detailed analysis. N2N4M is self-supervised and critically does not require zero-noise ground truth data, which is rarely available in Planetary Science applications. For more details please check our paper out: https://lnkd.in/dZPvseTu Grateful as always for the support of my co-authors and supervisors Rossella Arcucci and Cedric John, and can't wait to see what this method reveals about Martian geology! #AI4science #Hyperspectral #RemoteSensing #machinelearning #PlanetaryScience
Noise2Noise Denoising of CRISM Hyperspectral Data
arxiv.org
To view or add a comment, sign in
-
Enhance your legacy data with Machine Learning in OpendTect We are excited to announce a new paper by Carlos José Araque Pérez, Teresa Teixidó, Flor de Lis Mancilla, and José Morales, showcasing the use of OpendTect Machine Learning to enhance legacy data from the Granada Basin in Spain. The process of recovering and enhancing paper copy seismic sections involved three key phases: High-resolution scanning of paper copies to TIFF images and conversion to SEG-Y format. Post-stack processing to improve the resolution and lateral coherency of seismic events. Application of an OpendTect machine learning workflow to boost spatial resolution and improve signal-to-noise ratio and coherency. The authors report that these advancements have significantly improved the recovery and interpretation of seismic sections, leading to new geological insights into the complex Granada Basin. The paper, titled: “Reprocessing and Interpretation of Legacy Seismic Data Using Machine Learning from the Granada Basin, Spain,” is published in Tectonophysics and can be accessed here: https://lnkd.in/daXErRM3. dGB Earth Sciences #OpendTect #ML #MachineLearning #seismic #gesocience #geophysics
To view or add a comment, sign in
-
How #remotesensing and #sensorfusion technologies can provide valuable information for #greenhouse gas inventory and #biodiversity protection? Peatland is a type of wetland where dead plants, mainly mosses, accumulate over time and form a thick layer of organic material called peat. This peat builds up in waterlogged, low-oxygen conditions, preventing the plant material from fully decomposing. Peatlands are important for the environment and can perform a variety of important functions, such as filtering water, storing carbon, and providing habitat for wildlife. Therefore, it’s essential to monitor the natural peatland over time and detect its changes. We’ve proposed an encoder-decoder-based architecture for peatland classification that fuses two open-source satellite data, Sentinel-1 and Sentinel-2. We show the effect of fusion by comparing the multi-modal fusion architecture with uni-modals which are trained only based on one input data source. We also investigate the influence of skip connections as the main component of the encoder-decoder to recover fine-grained details that are lost during the downsampling process. All implementation and development stages have been done at Teknillinen tiedekunta – Faculty of Technology (University of Turku) in Turun yliopisto - University of Turku The experimental results are acquired on a study area in Finland which covers a variety of minerotrophic aapa mire peatlands. The dataset was collected by the Geological Survey of Finland (GTK) / Geologian tutkimuskeskus (GTK) to train the multi-modal fusion architecture for automatically labeling a pixel in a raster dataset to a specific category of peatlands. The paper was presented at the 27th International Conference on Information Fusion 2024 in Venice, Italy. I will share the article as soon as it is published and put its link in the comment. Lastly, I am so grateful for the opportunity to be a part of this wonderful project and work with like-minded, well-established, and highly talented scientists: Luca Zelioli, Fahimeh Farahnakian, Maarit Middleton, and Jukka Heikkonen. #informationfusion #datafusion #sensorfusion #artificialintelligence #machinelearning #deeplearning #datascience #monitoring #autonomoussystems #multimodal
To view or add a comment, sign in
-
Enhance your geological data analysis with ElasticDocs. 🌏 #ElasticDocs allows #engineers and #geoscientists to swiftly navigate thousands of report pages and identify #geological analogs using advanced search-and-cluster techniques. This #innovative solution is crafted by Iraya’s team of expert machine learning specialists and #geoscientists. #UnstructuredData #LargeLanguageModelling #ComputerVision #DataAtelier #earth #energy #environment #readtheearthbetter #IrayaEnergies #DataInsights #DataDrivenExploration #DataDrivenDiscovery #InnovateWithIraya #AIForEarthInsights #EarthInsightsAI
To view or add a comment, sign in
-
Enhancing ML models (through type, architecture, and hyperparameters) is not always the only approach for improving results, especially in the still-evolving field of earthquake prediction. In our recent publication, we focused on the data side to provide ML with more informative inputs. We introduced (a) event-based features and (b) new family features (clustering characteristics). This approach enhanced the input data by better reflecting the preparatory phase of labquakes. As a result, even a Random Forest model was able to forecast the likelihood of a large labquake in the next time window. I would like to thank all the support I received from all my coauthors specially Grzegorz Kwiatek . Check the full paper in the newly released JGR series "Journal of Geophysical Research: Machine Learning and Computation": https://lnkd.in/d-gQhh-b
Empowering Machine Learning Forecasting of Labquake Using Event‐Based Features and Clustering Characteristics
agupubs.onlinelibrary.wiley.com
To view or add a comment, sign in
-
🌊 Our recent paper on implementing the #Great_Lakes #Wave #Unstructured (#GLWUv2.0) forecast system at NOAA is now published in Geoscientific Model Development (#GMD)! 📝 During my time at #NOAA (just before I left), we devoted ourselves to bringing the state-of-the-art in #WAVEWATCH_III wave model, which we spent nearly 7 years enhancing for coastal capabilities and modernization, into operational use. The aim of this wave forecast system is to ensure the safety and efficiency of maritime activities in the Great Lakes region. Our paper delves into three key aspects of our work: 1- Operational Guidelines: We developed a robust framework for operationalizing models in highly restrictive environments, such as #NCO/#NCEP/#NWS, ensuring reliability even in the most critical forecasting situations. This framework encompasses aspects such as model reproducibility, simulation time, and coding standards. The outcome was a system that the operational center could operate without deep knowledge about wave models. 2- Scientific Advancements: Through rigorous scientific research, we enhanced the predictability and performance of the GLWUv2.0 model, overcoming numerous challenges posed by operational constraints. We demonstrated that the model is now capable of achieving a few-meter resolution in coastal areas without a penalty in runtime, thanks to the modernization efforts we implemented in WAVEWATCH III. 3- Future Directions: Looking ahead, we discuss the next steps for wave modeling, emphasizing how to improve wave modes across different scales, from global to coastal scales, and the need for wave observations during ice seasons. We also highlight the significance of #coupling our model with other Earth system components like #circulation, #ice, and #hydrology for better forecasting. Our validation against in situ data, covering both summer and wintertime conditions, demonstrates the reliability of GLWUv2.0 in diverse scenarios, including ice season challenges. This article not only sheds light on our achievements but also underscores the ongoing commitment to advancing wave forecasting technology for safer maritime operations in enclosed basins like the Great Lakes. My heartfelt gratitude goes out to my colleagues and friends, Saeideh Banihashemi, Jose-Henrique Alves, Aron Roland, Tyler Hesser, Jane McKee Smith and @Mary Anderson Bryant, whose dedication and expertise were invaluable throughout this endeavor. The GLWUv2.0 is in operation since May 9th, 2023. #WaveModeling #GreatLakes #NOAA #ERDC #CHL #Science #Research #Forecasting #MaritimeSafety 🚢💡
Great Lakes wave forecast system on high-resolution unstructured meshes
gmd.copernicus.org
To view or add a comment, sign in
-
Mapping Geologist | Remote Sensing & GIS Specialist | PhD in Economic Geology | Passionate about Sustainability & GeoAI.
Key Highlights of My PhD Research on GIS and Remote Sensing for Geological and Mineral Prospectivity Mapping The thesis explores the transformative potential of remote sensing techniques and GIS in geological applications, with a focus on reducing the time and cost involved in obtaining valuable geological information. By integrating multiple data sources within a GIS framework, the research aims to enhance the efficiency of mineral exploration and prevent the loss of potential resources by applying modern scientific methodologies. The Galat Sufur area, located in the mineral-rich Arabian-Nubian Shield, was chosen as an ideal case study due to its tectonic and geological features, along with the availability of relevant geological data. 1. The study involved extensive digital image processing of optical and radar data, including Landsat 8, Sentinel-2, ASTER, and Sentinel-1. These techniques enabled the accurate delineation of lithological units, geological structures, and hydrothermal alteration zones. The results were validated through field inspections and petrographic analysis, leading to an updated geological map of the area. 2. Advanced remote sensing techniques, such as Spectral Angle Mapper (SAM), Matched Filtering (MF), and Mixture Tuned Matched Filtering (MTMF), were applied to ASTER shortwave data (SWIR) to map mineral abundance in the study area. 3. In addition, hyperspectral measurements from the TerraSpec Halo Mineral Identifier device (ASD) were analyzed using machine learning techniques, including SAM, Spectral Feature Fitting, and Background-Removed Correlation, to precisely identify minerals. These results were further corroborated by petrographic investigations, confirming the potential of hyperspectral remote sensing for accurate mineral identification. Subsequently, the integration of ASD hyperspectral data with ASTER multispectral data effectively enhanced the mapping of minerals in the study area. 4. Utilizing the capabilities of GIS, mineral prospectivity mapping was performed using 26 predictive layers, including lithology, geological structures, gold-associated minerals, altered minerals, and hydrothermal zones. 5. Both knowledge-driven models (Weighted Sum) and data-driven models (Forest-based Classification and Regression, Support Vector Machine, and GeoAI tools for Automated Machine Learning) were applied to generate four prospectivity maps for gold mineralization. 6. The models' performance was validated, and the prospectivity maps were evaluated and compared against known mine and prospect locations in the study area. #gis #GeoAI #geologicalmapping #prospectivitymapping #remotesensing #Multispectralanalysis #hyperspectralanalysis
To view or add a comment, sign in
-
#MostCited Landslide Susceptibility Mapping Based on Interpretable Machine Learning from the Perspective of Geomorphological Differentiation ✍️ by Deliang Sun, Danlu Chen, Jialan Zhang, Changlin Mi, Qingyu Gu, and Haijia Wen 👉 https://brnw.ch/21wKv21 #landslide #machinelearning #geomorphological
Landslide Susceptibility Mapping Based on Interpretable Machine Learning from the Perspective of Geomorphological Differentiation
mdpi.com
To view or add a comment, sign in
-
Dear colleagues, I am pleased to announce the publication of a paper written by J Andrés Christen, @Tago, and myself entitled "Modeling evolutionary power spectral density functions of strong earthquakes via copulas." The paper has recently been published in the journal Soil Dynamics and Earthquake Engineering. If you would like to read the article in question, you may access it via the following link https://lnkd.in/eU7QYaJi.
Modeling evolutionary power spectral density functions of strong earthquakes via copulas
sciencedirect.com
To view or add a comment, sign in
MS Student at Department of Geology, Remote sensing Analyst
4moCongratulations