Welcome to the 1st workshop on Complex Data Challenges in Earth Observation (CDCEO) 2021. This workshop is held as a satellite event at the 30th ACM International Conference on Information and Knowledge Management (CIKM). CIKM 2021 runs as a fully virtual conference.
The workshop focuses on advancing research in Earth observation (EO) by effectively interpreting the high-dimensional heterogeneous data obtained by high-resolution remote sensing technologies.
Big data accumulated by ground, aerial, and satellite-based remote sensors at an unprecedented scale and resolution invite the application of modern data-hungry machine learning (ML) methods.
The workshop aims to bring together researchers in the fields of remote sensing, geographic information systems, weather and climate modeling, computer vision, and others with a general interest in applying data-driven models in EO.
The workshop invites advanced applications and method development in image and signal processing, data fusion, feature extraction, meta learning, and many more.
The workshop topics include but are not limited to:
spatiotemporal data processing and analysis;
multi-resolution, multi-temporal, multi-sensor, and multi-modal data fusion;
ML for weather and climate research;
deep learning and its applications to e.g., semantic segmentation, scene classification, and feature extraction;
advanced applications of time-series data analysis, e.g., urban sprawl, deforestation, crop monitoring, weather forecasts;
feature extraction, feature selection, and dimensionality reduction;
meta learning, including transfer learning, few-shot learning, and active learning;
data acquisition and efficient pre-processing of diverse remote sensing measurements including:
passive sensor images (panchromatic, multispectral, and hyperspectral);
active sensor data (LiDAR, RADAR, and SAR);
integration and aggregation of complementary remote sensing measurements;
advances in signal processing with applications to, e.g., unmixing, denoising;
benchmark datasets with application to EO.
Nowcasting Sattelite Application Facility, Spanish Meteorological Agency, Spain
Application of Artificial Intelligence/Machine Learning techniques in Remote Sensing for Meteorology
Abstract:
Meteorology is a field that traditionally has had big data volumes. With the advent of meteorological satellites than routinely observe the Earth, these data volumes have increased considerably. This tendency will grow even more as the satellite technology to observe the Earth evolves. As an example, the new generation of geostationary satellites that EUMETSAT will launch in the near future known as Meteosat Third Generation (MTG) will have instruments with increased spatial, temporal and channel resolution. The Flexible Combined Imager (FCI) will measure in 16 spectral channels and between 0.5 and 2 km spatial resolution measuring every 2.5 to 10 minutes. The InfraRed Sounder (IRS) will measure with a spatial resolution of 4 km every 60 minutes at around 2000 different channels at different wavelengths in the infrared. The total data rate will be increased more than 10-fold with respect to the previous Meteosat generation (Meteosat Second Generation).
Given these huge data volumes in satellite meteorology, it is clear that this field can benefit greatly from Machine Learning/Artificial Intelligence (ML/AI) techniques. One clear application is the diagnosis of the current state of the atmosphere from a global perspective using these instrument in combination with ML/AI. Another obvious candidate is the forecast of the weather from a few minutes to hours of lead time (Nowcasting) to several days (Weather Forecasting).
Even though these applications of ML/AI are quite obvious, it is not straightforward to apply them in practice. It must always be considered that several factors need to exist in order for these techniques to be successful. On one hand, the information we are trying to extract should be there in the original data. A careful analysis of the input data needs to be made to verify this. On the other hand, the problem at hand must be mathematically tractable by these techniques. In this presentation examples of several applications that can or cannot succeed will be given.
Bio:
Science coordinator for the EUMETSAT Nowcasting Satellite Application Facility (NWC SAF) and the State Meteorological Agency of the Government of Spain AEMET. He has worked with EUMETSAT in Darmstadt specializing in atmospheric profile retrievals obtained from hyper-spectral sounding instruments, such as the Infrared Atmospheric Sounding Interferometer, and planning for the future geostationary hyperspectral sounding mission MTG-IRS. He has also worked as a state meteorologist at AEMET.
European Centre for Medium-Range Weather Forecasts (ECMWF)
Machine Learning for Weather and Climate Predictions
Abstract:
The talk provides an overview of the work on machine learning methods that is ongoing at the European Centre for Medium-Range Weather Forecasts, and outlines how machine learning, and in particular deep learning, could help to improve weather predictions in the coming years. The talk will name challenges for the use of machine learning and suggest developments (research/software/hardware) that should enable the community of Earth system modelling to make quick progress.
Bio:
Peter is the AI and Machine Learning Coordinator at ECMWF and holds a University Research Fellowship of the Royal Society that enables him to perform research towards the use of machine learning, high-performance computing, and reduced numerical precision in weather and climate predictions. Peter has also a strong interest in the quantification of uncertainty of predictions for chaotic systems. Peter is coordinator of the MAELSTROM EuroHPC-Join Undertaking project, work-package leader of the ESiWACE2 H2020 project, and Co-Pi of an US-INCITE grant to perform season-long, global, storm-resolving simulations. Before moving to ECMWF, Peter has written his PhD thesis at the Max Planck Institute for Meteorology and has worked as PostDoc with Tim Palmer at the University of Oxford.
Department of Geodesy and Geoinformation, Technical University Wien, Austria
Earth Observation Data Centre for Water Resources Monitoring, Austria
Advancing the Understanding of Spaceborne Radar Observations using Machine Learning and Physical Models
Abstract:
Earth observation satellites equipped with radar sensors have become indispensable for monitoring the land surface from local to global scales. In particular, the Sentinel-1 satellites series represents a breakthrough as it acquires new radar imagery with a spatial resolution of 20 m every few days. Since the launch of the first Sentinel-1 satellite in 2014, several Petabyte of data have already been collected. Given the large data volume and the difficulties in the interpretation of the radar data, machine learning (ML) techniques are increasingly used alongside physical models to derive geophysical data sets such as soil moisture, water bodies, or forest type. In this contribution, I review a number of studies that have used ML techniques such as Support Vector Regression, Gradient Boosted Regression Trees, or Long-Short Term Memory for the classification and interpretation of Sentinel-1 radar data. The success of all these studies was routed in the availability of high quality reference data for ML training and validation. The ML models often worked surprisingly well, however, sometimes for the wrong reasons. In addition, the limits of the applicability of the trained models are often not clear. Therefore, the usefulness of ML for turning radar measurements into higher-value data sets depends critically on the context and intended application of the data. Further improvements in our understanding of spaceborne radar data can be expected from the combination of physical models and ML approaches.
Bio:
Wolfgang Wagner received the Dipl.-Ing. (MSc) degree in physics and the Dr.techn. (PhD) degree in remote sensing from the Technische Universität Wien (TU Wien), Austria, in 1995 and 1999 respectively. In support of his master and PhD studies he received fellowships to carry out research at the University of Bern (1993), Atmospheric Environment Service Canada (1994), NASA Goddard Space Flight Centre (1995), European Space Agency (1996), and the Joint Research Centre of the European Commission (1996-1998). From 1999 to 2001 he was with the German Aerospace Agency (DLR). In 2001 he was appointed professor for remote sensing at the Institute of Photogrammetry and Remote Sensing of TU Wien. From 2006 to 2011 he was the head of the Institute of Photogrammetry and Remote Sensing, from 2012 to 2019 the head of the newly founded Department of Geodesy and Geoinformation, and since 2020 he serves as the dean of the Faculty for Mathematics and Geoinformation. Furthermore, he is co-founder of the EODC Earth Observation Data Centre, where he has worked part-time as senior scientist since December 2014. Since 2018 he has also been affiliated with the Global Change Research Institute of the Czech Academy of Sciences (CzechGlobe).
| Video | Title of the talk | Speaker | Affiliation |
Video | 1st session – November 1st, 12:00 – 1:15 PM | Chair: Pedram Ghamisi | ||
| Invited talk: Advancing the Understanding of Spaceborne Radar Observations using Machine Learning and Physical Models by Wolfgang Wanger | Wolfgang Wanger | Department of Geodesy and Geoinformation, TU Wien, Austria Earth Observation Data Centre for Water Resources Monitoring, Austria | |
| Towards Very-Low Latency Storm Nowcasting through AI-Based On-Board Satellite Data Processing by Robert Hinz | Robert Hinz | DEIMOS Space S.L.U, Spain | |
| Uncertainty-Aware Graph-Based Multimodal Remote Sensing Setection of Out-of-Distribution Samples by Iain Rolland | Iain Rolland | University of Cambridge, UK | |
| Towards Geographical Aware Neural Networks for Geospatial Vector Data: A Case Study on Land Use and Land Cover ClassificationA by Marvin Mc Cutchan | Marvin Mc Cutchan | Institute of Advanced Research in Artificial Intelligence, Austria TU Wien, Austria | |
| 13:15 – 13:30 | COFFEE BREAK (Join us at the IARAI topia) | ||
Video | 2nd session – November 1st, 1:30 – 2:45 PM | ||
| 13:30 | Invited talk: Machine Learning for Weather and Climate Predictions by Peter Dueben | Peter Dueben | European Centre for Medium-Range Weather Forecasts (ECMWF) |
| 14:00 | Region-Growing Fully Convolutional Networks for Hyperspectral Image Classification with Point-Level Supervision by Yonghao Xu | Yonghao Xu | Wuhan University, China |
| 14:15 | Hyperspectral Anomaly Detection Based on Low-Rank Structure Exploration by Shizhen Cheng | Shizhen Cheng | Wuhan University, China |
| 14:30 | On the Exploitation of Heterophily in Graph-based Multimodal Remote Sensing Data Analysis by Catherine Taelman | TBA | TBA |
| 14:45 – 15:00 | COFFEE BREAK (Join us at the IARAI topia) | ||
| Poster Session – November 1st, 3:00 – 4:00 PM | |||
| Graph Neural Sparsification for Hyperspectral Image Classification with Local and Global Consistency by Haojie Hu | Haojie Hu | Xi’an Research Institute of High Technology, China | |
| Video | Hyperspectral Denoising: From Conventional techniques Towards Deep Learning ones by Behnood Rasti | Behnood Rasti | Alexander von Humboldt Research Fellow, Helmholtz Institute Freiberg for Resource Technology, Germany |
| Video | Point-based Weakly Supervised Deep Learning for Water Extraction from High-resolution Remote Sensing Imageryby Ming Lu | Ming Lu | Hunan University, China |
| Video | Machine Learning Model Development for Space Weather Forecast by Randa Natras | Randa Natras | Technical University of Munich, Germany |
| Large-Scale Hyperspectral Image Clustering Using Contrastive Learning by Yaoming Cai | TBA | TBA | |
| Video | FORCE on Nextflow: Scalable Analysis of Earth Observation data on Commodity Clusters by Fabian Lehmann | Fabian Lehmann | Institute for Computer Science, Humboldt-Universität zu Berlin, Germany |
| Change Detection for Hyperspectral Imagery based on Multi-layer Cascade Screening Strategy by Lian Liu | Lian Liu | Aerospace Information Research Institute, Chinese Academy of Sciences, China | |
| Video | End-to-end CNN-CRFs for Multi-date Crop Classification Using Multitemporal Remote Sensing Image Sequences by Laura Elena Cué La Rosa | Laura Elena Cué La Rosa | Pontifical Catholic University of Rio de Janeiro (PUC-Rio), Brazil |
| COFFEE BREAK (Join us at the IARAI) | |||
| Special Weather4cast Competition Session – November 1st, 4:15 – 6:00 PM | Chair: David Kreil | |||
| Invited talk: Application of Artificial Intelligence/Machine Learning techniques in Remote Sensing for Meteorology by Xavier Calbet | Xavier Calbet | Spanish Meteorological Agency AEMET, Spain | |
| The Weather4cast Stage1 Competition Design and Data by Pedro Herruzo | TBA | Institute of Advanced Research in Artificial Intelligence, Austria | |
| 1st prize: Spatiotemporal Weather Data Predictions with Shortcut Recurrent-Convolutional Networks: A Solution for the Weather4cast challenge by Jussi Leinonen | Jussi Leinonen | Federal Office of Meteorology and Climatology MeteoSwiss, Switzerland | |
| 2nd prize: Utilizing UNet for the future weather prediction: Weather4cast 2021’ by Sungbin Choi | Sungbin Choi | ||
| 3rd prize: A Variational U-Net for Weather Forecasting by Pak Hay Kwok, Qi Qi | Pak Hay Kwok, Qi Qi | ||
| Efficient Spatio-temporal Weather Forecasting with Deep Neural Networks by Akshay Punjabi, Pablo Izquierdo Ayala | Akshay Punjabi, Pablo Izquierdo Ayala | ||
| Spatiotemporal Swin-Transformer Network for Short Time Weather Forecasting by Alabi Bojesomo | Alabi Bojesomo | Khalifa University, UAE | |
| Closing remarks by Pedram Ghamisi | Pedram Ghamisi | Institute of Advanced Research in Artificial Intelligence, Austria | |
| Get-together (Join us at the IARAI topia) | |||
Authors are invited to submit original papers presenting research, position papers or papers presenting research in progress that have not been previously published, and are not being considered for publication elsewhere.
Blind reviewing process preformed by members of Program Committee will be applied to select papers based on their novelty, technical quality, potential impact, clarity, and reproducibility.
Workshop papers will be included in a CIKM companion volume published by http://ceur-ws.org/. Papers must be formatted in CEUR style guidelines and be submitted via EasyChair. The page limit is 4 – 6 pages plus references. At least one of the authors of the accepted papers must register for the workshop for the paper to be included into the workshop proceedings.
Special issue
Authors of the accepted papers will be invited to extend their work and submit it for a special issue of the IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing journal.
A special session of the workshop will present the winning solutions and highlights from a unique multi-sensor weather forecasting competition. Join the competition via weather4cast.ai and predict weather products in various Earth regions!
Weather forecasts are of obvious immediate value, but also are an important part of EO, informing about continuing changes of our environment. Modern ML methods have recently become viable alternatives to long standing physics-based forecasting solutions.
The goal of the competition is a short-term prediction of the selected weather products based on meteorological satellites data obtained in collaboration with AEMET/ NWC SAF. The competition data are presented in a form of weather movies that consist of multi-channel images encoding the cloud properties, temperature, turbulence, and rainfall. The images are recorded at 15 minute intervals through the entire year in various Earth regions. Learn more…
Institute of Advanced Research
in Artificial Intelligence, Austria
Helmholtz-Zentrum Dresden-Rossendorf, Germany
University of Extremadura, Spain
Wuhan University, China
Institute of Advanced Research in Artificial Intelligence, Austria
Silesian University of Technology, Poland
Institute of Advanced Research in Artificial Intelligence, Austria
Spanish Meteorological Agency, Spain
European Space Agency
Φ-lab, Italy
Earth Observation Data Centre for Water Resources Monitoring, Austria
Institute of Advanced Research in Artificial Intelligence, Austria
Helmholtz-Zentrum Dresden-Rossendorf, Germany
Institute of Advanced Research in Artificial Intelligence, Austria
Institute of Advanced Research in Artificial Intelligence, Austria
European Space Agency
Φ-lab, Italy
Spanish Meteorological Agency, Spain
Earth Observation Data Centre for Water Resources Monitoring, Austria
cdceo@iarai.ac.at