|
AUTOMATED IDENTIFYING OF ENVIRONMENTAL CHANGES USING SATELLITE TIME-SERIES
|
Copernicus programme provide a huge amount of image data for many applications around the globe. Besides data being free and open, Sentinels also holds great potential to use dense observation time-series in various dynamical Earth environment [...] |
Dr. Mika Karjalainen, The Finnish Geospatial Research Institute, FINLAND |
Prof.. Yan Song, China University of Geoscience (Wuhan), CHINA |
Data Analysis |
Copernicus programme provide a huge amount of image data for many applications around the globe. Besides data being free and open, Sentinels also holds great potential to use dense observation time-series in various dynamical Earth environment monitoring applications. The amount of EO data is extensive, therefore it is of high importance to develop automated tools to process and analyse the data. In the FGI, we have developed a toolkit (EODIE) to automatically process Sentinel satellite images, to extract time-series data (NDVI or other index, or SAR backscattering values). When the EO time-series data are combined with ancillary information (training data set), it is possible to design intelligent classification tools to automatically identify causes for environmental changes. Our objective is to use machine learning techniques, Random forest and Neural Networks based methods, and design a classification tool to be used in automatic identification of environmental changes. We aim to test the classification toolkit in the field of agriculture (crop species classification in Finland), forestry (forest thinning mapping in Finland), and coastal area monitoring (China). The proposed research connects the complementary experiences of the European and Chinese partners to develop innovative techniques for analysing EO time-series data. The research directly aims at advancing the scientific careers of young researchers involved with the project by providing good opportunities to write articles for top journals in the field of remote sensing. For example, the use of EODIE toolkit (Earth Observation Data Information Extractor) is already applied in the demonstration projects in agricultural mapping in Finland, and likely of interest to the scientific community in Dragon 5 context. The proposed research is partially funded by the organizations of the reseach teams and other outside funded projects.
|
|
BIG DATA INTELLIGENT MINING AND COUPLING ANALYSIS OF EDDY AND CYCLONE
|
We here propose a new algorithm for parallel identification of mesoscale eddies from global satellite altimetry data. And a new hybrid mesoscale eddy tracking method to enhance the eddy tracking accuracy from global satellite altimeter data.We [...] |
Dr. Wang Shuai, Imperial College London, UK |
Dr. Fenglin Tian, Ocean University of China, CHINA |
Data Analysis |
We here propose a new algorithm for parallel identification of mesoscale eddies from global satellite altimetry data. And a new hybrid mesoscale eddy tracking method to enhance the eddy tracking accuracy from global satellite altimeter data.We will build a tropical cyclone data set globally based on a state-of-the-art atmospheric reanalysis product. Combined with the long time eddy identification data products, we intend to calculate the probability of the encounter between cyclones and mesoscale eddies.
|
|
LARGE-SCALE SPATIAL-TEMPORAL ANALYSIS FOR DENSE SATELLITE IMAGE SERIES WITH DEEP LEARNING
|
The latest decade has witnessed a great development of satellite remote sensing sensors, Earth Observation (EO) Big data with huge quantities of satellite images with high spatial and temporal resolution are available. The successive observation [...] |
Prof.. Daniela Faur, University Politehnica of Bucharest, ROMANIA |
Dr. Weiwei Guo, Tongji Universtiy, CHINA |
Data Analysis |
The latest decade has witnessed a great development of satellite remote sensing sensors, Earth Observation (EO) Big data with huge quantities of satellite images with high spatial and temporal resolution are available. The successive observation for long periods of time of a large area with short revisit interval from space is resulting in dense satellite image time series (SITS). SITS are a new type of EO product allowing us to analyze and mine not only spatial but spatio-temporal dynamic evolution information content about the scene structure and objects. This need to exploit spatial and temporal information contents from SITS is increasing with a wide range of applications, including urban development, target dynamic monitoring, precise agriculture, forest, and etc. Meanwhile, the open and free data access from many EO missions as in the Copernicus program or ChinaÔÇôBrazil Earth Resources Satellite 4 opens new fantastic perspectives for research and applications of Artificial Intelligence for EO A4EO. With a significant progress of AI, especially the major breakthrough of deep learning has proven to be an extremely powerful tool in many fields including computer vision, speech recognition, natural language processing, etc, and has also been enjoyed into the geoscience and remote sensing community for remote sensing big data analysis. However, contrary to multimedia image analysis which has been boosted by advanced deep neural networks and easily available big multimedia training data, the deep learning techniques and automatic tools for scale dense SITS analysis are limited, and large-scale benchmark dataset is not exist. Moreover, dense SITS analysis raises some specific challenges. Novel network architectures and datasets have to be developed exploiting the temporal information jointly with the spatial and spectral information of the data, but cope with multi-modal, multi resolutions and multi sensors data in a synergistic way. In this project, we intend to develop hybrid explainable AI, deep learning and big data analytics techniques and tools for large-scale dense SITS analysis and address the challenges to advance the state of the art in this area. The overall objective of this project is to provide an effective solution for large-scale dense SITS analysis, being capable of automatic discovery of regularities, relationships, and dynamic evolution, leading to a better and easier understanding of the underlying processes of specific scenes and targets. Specifically, the objectives are:(1) Develop weakly supervised deep learning techniques for object extraction and semantic classification for remote sensing images. (2) Develop deep spatial-temporal network techniques for large dense SITS clustering, classification and prediction. (3) Exploit deep change detection techniques for multi-temporal satellite images. We will compile a large-scale spatial-temporal data sensor for spatial-temporal change detection both for optical remote sensing images and for SAR images and develop novel deep neural network architectures and learning techniques for our specific data tensor and change detection task. (4) Develop spatial-temporal fusion and synergic computation techniques of Multi-modal, Multi-resolutions and Multi-sensor images for SITS mining, classification, and change analysis. The scientists of CEOSpaceTech – the research center within Politehnica University Bucharest – Romania, Tongji University and Shanghai Jiaotong University – China will tightly collaborate to advance these innovative techniques. For the evaluation and validation process we consider 2 use cases, targeting areas of Romania – EU for ecosystem monitoring of an UNESCO protected area and Shanghai – China for urban evolution in support of smart and sustainable urban information services. A list of related funding projects able to co-financing this project is available in the Annexes.
|
|
RESEARCH AND APPLICATION OF DEEP LEARNING FOR THE IMPROVEMENT AND ASSIMILATION OF SIGNIFICANT WAVE HEIGHT AND DIRECTIONAL WAVE SPECTRA FROM MULTI-MISSIONS
|
Remotely sensed ocean waves from European and Chinese space missions have significantly supplemented the insufficient coverage of traditional wave observations such as buoys. The accuracy of the wave remote sensing is not only important for the [...] |
Dr. Lotfi Aouf, Meteo-France, Division Marine et Oceanographie, FRANCE |
Dr. Jiuke Wang, National Marine Environmental Forecasting Center (NMEFC) , CHINA |
Data Analysis |
Remotely sensed ocean waves from European and Chinese space missions have significantly supplemented the insufficient coverage of traditional wave observations such as buoys. The accuracy of the wave remote sensing is not only important for the wave forecast, but also critical to the air-sea interactions, which impact significantly weather and climate projections. Therefore, it is of great importance to improve the accuracy of retrieved wave products before their usage. The assimilation technique is known by its efficiency to improve the forecast accuracy by using the observations. Beside of good quality observations, a proper method of assimilation is also needed to make the best use of the observations, which will greatly impact the assimilation effects. Deep learning, which is based on artificial neural networks, has proved its efficiency and effectiveness in computer vision, speech recognition and many fields. It is innovative to apply this powerful method into both improvement and assimilation of wave products retrieved from several sensors such as SAR, wave spectrometer and altimeters available on European and Chinese satellites. Based on the satellite missions of Sentinel-1A/B SAR (Synthetic Aperture Radar), ENVISAT ASAR, Sentinel-3A/B altimetry, HY-2A/B altimetry, CFOSAT wave spectrometer SWIM (Chinese-French Oceanic Satellite, Surface Waves Investigation and Monitoring), this proposal will focus on:1) Finding the key factors which affect the accuracy through the detailed assessments of wave observations from each mission. This is critical to decide which parameter should go into the deep learning network. 2) Designing and building Deep Neural Networks (DNN) to improve the accuracy of altimetry missions (delay Doppler altimetry mode) by doing related researches and experiments. The structure and hyper-parameter will be carefully tuned and optimized to obtain a robust correction model. 3) Developing a model to correct the spectra from SAR and SWIM missions based on deep learning. Processing the wave directional spectrum as an image, techniques such as CNN (Convolutional Neural Network) will be applied in the correction. 4) Bringing out and verifying a new method of assimilation by combining techniques from deep learning, including those used for the quality control procedure before the assimilation and the method of data merging.5) Reprocessing the data from multi-missions using the deep learning correction. By using the corrected wave data in the assimilation, a global wave reanalysis from the model MFWAM will be produced and will highlight the use of the new deep learning assimilation method.6) Providing a better estimate of Modulation Transfer Function (MTF) of SAR and SWIM by using deep learning technique from model wave spectra. To demonstrate the feasibility of these objectives above, a series of exploratory experiments have been done by our teams. According to the comparisons between the corrections and buoy observations, the DNN has removed 77% of bias and improve 22% of SI (Scatter Index) for the SWH (Significant Wave Height) from Sentinel-3A. The DNN also removes 98% of bias and 26% of SI for the SWH of CFOSAT SWIM Nadir observation. The directional spectra from CFOSAT SWIM can be also improved by the combination of CNN and DNN. The deep learning can total decrease 45% of bias and lower the SI by 35%. Therefore, the feasibility experiment results have proved that deep learning is very promising when applied in the improvement of wave products retrieved from different sensors such as SWIM, SAR, and altimeters. So it should now be extended to other European and Chinese satellite missions. Furthermore, the study by using the deep learning in assimilation system will also be investigated. This proposal is funded by National Key R&D Program of China on Global high-quality marine data assimilation research and operational application and other relevant programs.
|
