Artificial Intelligence applications in natural hazards research
The significant increase in computational power has led to an explosion of artificial intelligence applications, in an enormous number of sciences, especially in the field of geosciences. Machine learning, deep learning and other statistical and non-statistical methods are already successfully applied for detecting, mapping and monitoring natural hazards’ evolution across space and time.
The session is focused on applications of advanced CNNs (deep learning algorithms) for detecting and mapping landslide features at very detailed scales with minimum effort. Particular focus is given to deep learning models capable of detecting and mapping landslides and landslides features from various platforms and sensors; the prediction of landslide hazards using recent artificial intelligence models, like long-short-term memory; pre-trained models used for landslide detections from optical and radar imagery.
- Machine-learning methods applied to landslide susceptibility and hazard modelling using remote-sensed data
- Deep learning applications in landslide research
- Pretrained models for landslide detection and mapping
- Self-learning AI models
- Space-time landslide prediction through LSTM models
Monitoring and mapping multi-hazards under climate change
Understanding Earth’s system natural processes, especially in the context of global climate change, has been recognised globally as a very urgent and central research direction which needs further exploration.
The availability of over 50 years of image collection by satellites, as Landsat mission offers, combined with other satellite imagery archives (SPOT, CORONA etc.), opens new doors for building highly accurate inventories of natural hazards that occurred in the past. These highly accurate inventories will positively impact the accuracy of future predictions across time and space, thus helping the government take optimal decisions regarding multi-hazards and risk mitigations. Furthermore, as these archives are constantly growing with imagery collected from recent satellite missions, building and updating these inventories is expected to shift from a semi-automated one towards an automated one.
The session focuses on gathering scientific researchers related to this topic, aiming to highlight ongoing research and new applications in the field of satellite and aerial time-series imagery applications for natural hazards monitoring, mapping and modelling under the current climate change. It is open to satellite time-series processing algorithms and applications applied to different types of remote sensing data for investigating longtime processes. Both applied and theoretical research contributions focused on novel methods and applications of satellite and aerial time-series imagery, data acquired in all regions of the electromagnetic spectrum are welcomed.
- Satellite time-series application for building natural hazards inventories
- Sparsed aerial imagery – analogue and digital – for multi-temporal inventories
- Uncertainty estimations in predicting natural hazards from earth observation data
- Online services for landslide monitoring
- Climate change impact on the multi-hazards occurrence
Recent Earth Observation technology applications in natural hazards research
With the launch of new satellite platforms with a high revisit time, combined with the increasing capability for collecting repetitive ultra-high aerial images through unmade aerial vehicles, the scientific community have new opportunities for developing and applying new image processing algorithms to solve old and new environmental issues.s.
The high amount of earth observation data, combined with the constant increase in computational power, and the latest technology in the field of artificial intelligence, image recognition and advanced statistical analyses, made the monitoring of natural hazards evolution across space and time easily reached, even by non-specialists. Near real-time systems based on video feeds (ground and aerial platforms), high-resolution imagery, and point-cloud datasets are among the most used in geoscience.
The session focused on satellite and aerial imagery (UAV or other platforms) and other ground sensors to map and monitor geomorphological, geological, and hydrological processes in near-real or short revisits. Case studies from different environments and applications are welcomed.
- Near real-time monitoring of natural processes
- UAV applications for mapping natural hazards
- Satellite time-series applications for natural hazard monitoring
- Sensors fusion applications in natural hazards research
- Software tools that exploit recent technologies and algorithms in image processing