Blog – Analysis and expert opinion 5 MIN. READ SHARE Project Description GEMA is an 18-month project (August 2023 – February 2025) that aims to develop innovative solutions by integrating artificial intelligence in the interpretation of geophysical data. Its main objective is the creation of models based on deep neural networks that combine multiple data sources, such as geophysics (magnetic, radiometric and hyperspectral), satellite imagery and digital terrain models. These models will allow the generation of ultra-high-resolution maps, improving the accuracy and speed of geological interpretation. The project aims to reduce the time and costs associated with the manual interpretation of large volumes of data, improving the efficiency of exploration processes. GEMA is led by Xcalibur and is part of a wider initiative focused on the development of artificial intelligence applied to geophysics. Project Results and Milestones Achieved Optimisation of Geophysical Data using Tie Level (Tabular Structure) The Tie Level tool has optimised the correction of artefacts in geophysical data through an approach based on Bayesian models and spatial RBF (Radial Basis Function Interpolation). Its implementation has significantly reduced the need for manual adjustments, providing more homogeneous and accurate results in less time. Milestones achieved: More than 250,000 geophysical data points in the study regions have been corrected, reducing processing time by 40 % compared to traditional methods. The accuracy of the interpolation models was improved by 15%, increasing the reliability of the data used in exploration. Geographical scope of implementation: Finland Mongolia France Detection of Anomalies in Magnetic Lineaments with Computer Vision (CNN Lineaments) The CNN Lineaments tool has been developed for the automatic detection of geological structures using Computer Vision techniques and segmentation of geophysical images. Its implementation has optimised the interpretation of magnetic lineaments, facilitating the identification of structural anomalies. Milestones achieved: Identification of more than 1,200 magnetic lineaments in the study areas, with a 35 % reduction in the need for manual corrections. Successful application of the model in Nigeria and La Guajira (Colombia), where its effectiveness has been validated with 90 % accuracy in anomaly detection. Geographical scope of implementation: Karamoja, Uganda Katanga, Democratic Republic of Congo (DRC) Nigeria (model validation) La Guajira, Colombia (model validation). Magnetic Variation Correction in Directional Drilling Magnetic variation correction tools have been developed to improve the accurate positioning of boreholes in exploration and resource extraction. This has optimised borehole orientation, reducing errors and increasing operational efficiency. In addition, work has been done on estimating electromagnetic fields in locations without measurement stations, facilitating data acquisition in remote areas. Milestones achieved: Improved borehole positioning accuracy by 25 %, reducing borehole trajectory errors. Generation of an electromagnetic field estimation model that has enabled data collection in more than 20 locations without measurement stations in Canada. Geographical scope of implementation: The model has been trained and validated using data from the following regions in Canada, where Directional Drilling activity requires high accuracy: Alberta (majority of validation stations). British Columbia (Wells Gray) Saskatchewan (Weyburn, Rabbit Lake) Northwestern Territories (Fort Simpson, Fort Smith) Funding Total EU funding allocated: 1,206,071.80 €. Total project budget: 1,418,908.00 €. < SEE PREVIOUS NEW SEE NEXT NEW > < > Related news Blog – Analysis and expert opinion Startup Of The Week: Xcalibur Smart Mapping SEE MORE > Blog – Analysis and expert opinion Low-Level Flights to Image Geology Over Parts of Louisiana, Mississippi, and Illinois | U.S. Geological Survey SEE MORE > Blog – Analysis and expert opinion Reducing CO2 emissions from deforestation and forest degradation: understanding REDD+ SEE MORE > Announcements Blog – Analysis and expert opinion The power of remote sensing in monitoring land use changes SEE MORE > Blog – Analysis and expert opinion CRMA: why Europe’s raw materials revolution needs Smart data SEE MORE > Blog – Analysis and expert opinion Outlining the danger: the vital role of risk maps in the battle against fires SEE MORE > Blog – Analysis and expert opinion Unveiling advanced techniques: A-DInSAR for terrain deformation detection SEE MORE > Contact us. FILL THE FORM
SHARE Project Description GEMA is an 18-month project (August 2023 – February 2025) that aims to develop innovative solutions by integrating artificial intelligence in the interpretation of geophysical data. Its main objective is the creation of models based on deep neural networks that combine multiple data sources, such as geophysics (magnetic, radiometric and hyperspectral), satellite imagery and digital terrain models. These models will allow the generation of ultra-high-resolution maps, improving the accuracy and speed of geological interpretation. The project aims to reduce the time and costs associated with the manual interpretation of large volumes of data, improving the efficiency of exploration processes. GEMA is led by Xcalibur and is part of a wider initiative focused on the development of artificial intelligence applied to geophysics. Project Results and Milestones Achieved Optimisation of Geophysical Data using Tie Level (Tabular Structure) The Tie Level tool has optimised the correction of artefacts in geophysical data through an approach based on Bayesian models and spatial RBF (Radial Basis Function Interpolation). Its implementation has significantly reduced the need for manual adjustments, providing more homogeneous and accurate results in less time. Milestones achieved: More than 250,000 geophysical data points in the study regions have been corrected, reducing processing time by 40 % compared to traditional methods. The accuracy of the interpolation models was improved by 15%, increasing the reliability of the data used in exploration. Geographical scope of implementation: Finland Mongolia France Detection of Anomalies in Magnetic Lineaments with Computer Vision (CNN Lineaments) The CNN Lineaments tool has been developed for the automatic detection of geological structures using Computer Vision techniques and segmentation of geophysical images. Its implementation has optimised the interpretation of magnetic lineaments, facilitating the identification of structural anomalies. Milestones achieved: Identification of more than 1,200 magnetic lineaments in the study areas, with a 35 % reduction in the need for manual corrections. Successful application of the model in Nigeria and La Guajira (Colombia), where its effectiveness has been validated with 90 % accuracy in anomaly detection. Geographical scope of implementation: Karamoja, Uganda Katanga, Democratic Republic of Congo (DRC) Nigeria (model validation) La Guajira, Colombia (model validation). Magnetic Variation Correction in Directional Drilling Magnetic variation correction tools have been developed to improve the accurate positioning of boreholes in exploration and resource extraction. This has optimised borehole orientation, reducing errors and increasing operational efficiency. In addition, work has been done on estimating electromagnetic fields in locations without measurement stations, facilitating data acquisition in remote areas. Milestones achieved: Improved borehole positioning accuracy by 25 %, reducing borehole trajectory errors. Generation of an electromagnetic field estimation model that has enabled data collection in more than 20 locations without measurement stations in Canada. Geographical scope of implementation: The model has been trained and validated using data from the following regions in Canada, where Directional Drilling activity requires high accuracy: Alberta (majority of validation stations). British Columbia (Wells Gray) Saskatchewan (Weyburn, Rabbit Lake) Northwestern Territories (Fort Simpson, Fort Smith) Funding Total EU funding allocated: 1,206,071.80 €. Total project budget: 1,418,908.00 €.
Project Description GEMA is an 18-month project (August 2023 – February 2025) that aims to develop innovative solutions by integrating artificial intelligence in the interpretation of geophysical data. Its main objective is the creation of models based on deep neural networks that combine multiple data sources, such as geophysics (magnetic, radiometric and hyperspectral), satellite imagery and digital terrain models. These models will allow the generation of ultra-high-resolution maps, improving the accuracy and speed of geological interpretation. The project aims to reduce the time and costs associated with the manual interpretation of large volumes of data, improving the efficiency of exploration processes. GEMA is led by Xcalibur and is part of a wider initiative focused on the development of artificial intelligence applied to geophysics. Project Results and Milestones Achieved Optimisation of Geophysical Data using Tie Level (Tabular Structure) The Tie Level tool has optimised the correction of artefacts in geophysical data through an approach based on Bayesian models and spatial RBF (Radial Basis Function Interpolation). Its implementation has significantly reduced the need for manual adjustments, providing more homogeneous and accurate results in less time. Milestones achieved: More than 250,000 geophysical data points in the study regions have been corrected, reducing processing time by 40 % compared to traditional methods. The accuracy of the interpolation models was improved by 15%, increasing the reliability of the data used in exploration. Geographical scope of implementation: Finland Mongolia France Detection of Anomalies in Magnetic Lineaments with Computer Vision (CNN Lineaments) The CNN Lineaments tool has been developed for the automatic detection of geological structures using Computer Vision techniques and segmentation of geophysical images. Its implementation has optimised the interpretation of magnetic lineaments, facilitating the identification of structural anomalies. Milestones achieved: Identification of more than 1,200 magnetic lineaments in the study areas, with a 35 % reduction in the need for manual corrections. Successful application of the model in Nigeria and La Guajira (Colombia), where its effectiveness has been validated with 90 % accuracy in anomaly detection. Geographical scope of implementation: Karamoja, Uganda Katanga, Democratic Republic of Congo (DRC) Nigeria (model validation) La Guajira, Colombia (model validation). Magnetic Variation Correction in Directional Drilling Magnetic variation correction tools have been developed to improve the accurate positioning of boreholes in exploration and resource extraction. This has optimised borehole orientation, reducing errors and increasing operational efficiency. In addition, work has been done on estimating electromagnetic fields in locations without measurement stations, facilitating data acquisition in remote areas. Milestones achieved: Improved borehole positioning accuracy by 25 %, reducing borehole trajectory errors. Generation of an electromagnetic field estimation model that has enabled data collection in more than 20 locations without measurement stations in Canada. Geographical scope of implementation: The model has been trained and validated using data from the following regions in Canada, where Directional Drilling activity requires high accuracy: Alberta (majority of validation stations). British Columbia (Wells Gray) Saskatchewan (Weyburn, Rabbit Lake) Northwestern Territories (Fort Simpson, Fort Smith) Funding Total EU funding allocated: 1,206,071.80 €. Total project budget: 1,418,908.00 €.