Final Report Summary - GIONET (GMES Initial Operations – Network for Earth Observation Research Training)
GIONET "GMES Initial Operations - Network for Earth Observation Research Training" is a European Centre of Excellence for Earth Observation Research Training for Early Stage Researchers underpinning the Copernicus (Global Monitoring for Environment and Security, formerly GMES) land monitoring and emergency monitoring core services. Copernicus is Europe’s flagship operational Earth Observation programme. GIONET is a partnership of leading Universities, research institutes and private companies across Europe and employed 14 early stage researchers for three-year doctoral projects undertaken in collaboration between the academic and private sector. The researchers undertook secondments in industry and universities across Europe, as well as receiving the best technical training and scientific education. This training was delivered through individually supervised research, four international summer schools and local training. It was implemented through supervised research in five thematic areas: forest monitoring, land cover change, coastal zone and freshwater monitoring, geohazards and emergency response, and climate adaptation and emergency response.
GIONET was a companion for the European Copernicus programme during its GMES Initial Operations (GIO) phase from 2011-2013 and beyond. GIONET has trained 14 highly skilled researchers for the operational phase of the Copernicus land monitoring and emergency services by providing postgraduate training in Earth Observation Science including complementary skills and work experience in the private and academic sector and developing a collaborative training network with secondments to other GIONET organisations.
The objectives of GIONET were to:
1. Provide broad postgraduate training in Earth Observation Science that exposes students to different research disciplines and complementary skills, providing work experiences in the private and academic sectors, leading to a recognized qualification (Doctorate).
2. Enable access to first class training in both fundamental and applied research skills to early-stage researchers at world-class academic centres and market leaders in the private sector.
3. Build on the experience gained from previous Copernicus/GMES Research and Development projects in the land monitoring and emergency services areas, including FP5 BIOPRESS, FP6 Geoland and BOSS4GMES, FP7 Geoland2, SAFER and RESPOND.
4. Develop a collaborative training network, through the placement of researchers supported by the grant for short periods (3 months) in other GIONET organisations, and by hosting nationally funded students on request.
5. Provide a flexible training environment with personal career development planning, that early stage researchers can adapt to their cultural, disciplinary and gender needs and in respect of their national education systems.
Since the beginning of the project in 2011, GIONET has accomplished many scientific breakthroughs and technological innovations. In the field of forest monitoring, GIONET produced an algorithm that improved the forest monitoring of the Congo basin using Synthetic Aperture Radar (SAR). James Wheeler at the University of Leicester analysed a time series of 12 ALOS-PALSAR scenes in the Democratic Republic of Congo. He mapped the entire Congo Basin and produced yearly forest cover maps of a five degree by five degree region. He was invited to the 21st Science Team meeting of the Kyoto and Carbon Initiative held in Kyoto, Japan in 2014. Pedro Rodriguez Veiga at the University of Leicester investigated a data mining approach based on the Maximum Entropy algorithm to map forest aboveground biomass in Mexico, Siberia and Spain, with spatial uncertainty and forest probability maps. He demonstrated that optical, SAR and elevation data contain different information that can be combined by this approach to produce better estimations than with a single sensor approach. This method is now being studied in the European Space Agency project GLOBBIOMASS for global application. Martyna Stelmaszczuk-Górska at the University in Jena developed a large-scale forest monitoring scheme and evaluated different biomass retrieval models over Siberian boreal forests. The results are published in the peer-reviewed journal the Siberian Journal of Forest Science.
In the area of land cover and change, Bernard Spies at Airbus DS (formerly Astrium UK) used Random Forest Classifiers for TerraSAR-X (X-band), ENVISAT ASAR (C-band) and ALOS PALSAR (L-band). Two test sites were studied for wet and dry seasons, Cameroon-Dry, Cameroon-Wet, Tanzania-Dry and Tanzania-Wet. The dry season gave higher classification accuracies than the wet season. X-band SAR contributes the most to the classification accuracies of the forested site (Cameroon) and L-band has the highest contribution to the semi-arid (Tanzania) site. The best dual-frequency combination for both sites is L- and X-band. Shailesh Shrestha at IGIK developed a method for using adaptive neighbourhood information for land cover change detection. He used the concept of Geographic Image Object Analysis (GEOBIA) at very local scale in Warsaw, Poland using high-resolution QuickBird images. Multi-scale remote sensing data synergies for land process studies were studied by Maxim Chernetskiy at the University of Jena. His work improved the retrieval of land biophysical and spectral parameter retrieval from remote sensing, data assimilation and radiative transfer modelling.
GIONET advanced the application of remote sensing for coastal zone and freshwater monitoring. Sybrand van Beijma at Airbus DS used airborne radar for salt marsh vegetation mapping in Wales for the first time. His work now forms a reference methodology for the processing of data from the Airbus DS airborne SAR instrument, including the publically-funded AirSAR campaign in the UK in 2014. Dimitris Stratoulias at the Hungarian Academy of Sciences Centre for Ecological Research, Balaton Limnological Institute used airborne hyperspectral data for mapping emergent aquatic plants and studying reed die-back around the lake. The method can be applied to Sentinel-2 satellite data. Stephanie Palmer at the Balaton Limnological Institute explored the use of remote sensing to measure the phytoplankton biomass in the water of Lake Balaton, using a ten year time series of Medium Resolution Imaging Spectrometer (MERIS) satellite imagery, ship-mounted Light Detection and Ranging (LiDAR), and water sampling and laboratory measurements. She demonstrated the feasibility of retrieving accurate and quantitative information on Lake Balaton phytoplankton biomass dynamics through the use of remote sensing techniques.
In the field of geohazards and emergency response, Penelope Kourkouli at GAMMA Remote Sensing developed a hybrid method based on Differential Synthetic Aperture Radar Interferometry (DInSAR) and Persistent Scatterer Interferometry (PSI) and tested it in the Venice Lagoon, Italy and over a semi-arid terrain in Jordan. Jessica Papke at GAMMA investigated spaceborne interferometric Synthetic Aperture Radar (InSAR) data processing and terrestrial radar interferometry for the monitoring of landslide displacements. Topography and surface velocities were mapped for an Alpine study site in the Western Swiss Alps with satellite and ground-based radar interferometry. The surface velocity of an active rock glacier was also mapped with a ground-based radar interferometer. Christoff Fourie at the German Aerospace Center (DLR) developed new methods for sample supervised geographic object based image analysis to support emergency response actions. Two new approaches have been prototyped.
In support of climate adaptation and emergency response, Matthew Ofwono at IGIK applied remote sensing methods for the integration of long-term trend analysis in drought detection and monitoring of drought impacts on crop yield. Soil moisture deficit modelling was implemented in the Decision Support System for Agro-technology Transfer (DSSAT) to estimate wheat yield. This new approach has been demonstrated in incorporating to DSSAT model besides of soil moisture deficit index also fAPAR and LAI from Terra/MODIS satellite what significantly reduced uncertainty in wheat yield prediction. Linda Moser at the German Aerospace Center (DLR) monitored wetlands in semi-arid Africa with optical and SAR data over 15 years (2000-2014). Indicators for water stress and associated land use changes, potentially connected to human crisis situations, were deducted from remote sensing.
Expected final results and their potential impact and use
• Doctoral Training: All Early Stage Researchers have stayed in employment for the planned 36 months in total and are now in various stages of completing their PhD studies. Stephanie Palmer and Sybrand van Beijma have already passed their PhD examinations, Dimitris Stratoulias has submitted his PhD thesis and Matthew Ofwono and Shailesh Shresta have declared their intent to submit it within 3 months. All other ESRs are writing up their thesis aiming for completion by September 2015. Four GIONET summer schools were held to train the researchers and external participants.
• Highly Skilled Researchers: Many GIONET researchers have already secured research employment after the end of their Marie Curie fellowships, for example in the University of Twente (NL), University of Leicester (UK) and the private companies Gamma Remote Sensing (CH) and Geospatial Insight (UK).
• Research from GIONET has been published in 22 highly rated journals, including Remote Sensing of Environment, which is considered the top journal in the field of remote sensing (Impact Factor: 4.77). Other journals include: Applied Geography (Impact Factor 3.08) Remote Sensing (Impact Factor: 2.729) International Journal of Applied Earth Observation and Geoinformation (Impact Factor: 2.809) Engineering Geology for Society and Territory (Impact Factor: 0.721) Earth Science Informatics Impact Factor; 0.694 and Atmospheric and Oceanic Physics (Impact Factor 0.597).
• Networking and Academia-Industry Collaboration: Seven new approved associated partners have joined GIONET: National Observatory of Athens, Greece; University of Stirling, UK; Università degli Studi di Padova, Italy; Danko (Poland), EOS Jena (Germany), Envirosense (Hungary) and Blackbridge (Germany).
• Dissemination and Exploitation: A book contract has been signed with Wiley-Blackwell Publishers to disseminate the GIONET results and 14 book chapters have been submitted and are now undergoing peer-review: Balzter, H. (Ed., submitted): “Earth Observation for Land and Emergency Monitoring - Innovative concepts for environmental monitoring from space”. Wiley, London. Furthermore, GIONET methods and results are implemented and further developed in other projects, including the MaxEnt biomass mapping for GLOBBIOMASS, a reference methodology for the processing of Airbus DS airborne SAR data, and new contract work won by GAMMA.
GIONET was a companion for the European Copernicus programme during its GMES Initial Operations (GIO) phase from 2011-2013 and beyond. GIONET has trained 14 highly skilled researchers for the operational phase of the Copernicus land monitoring and emergency services by providing postgraduate training in Earth Observation Science including complementary skills and work experience in the private and academic sector and developing a collaborative training network with secondments to other GIONET organisations.
The objectives of GIONET were to:
1. Provide broad postgraduate training in Earth Observation Science that exposes students to different research disciplines and complementary skills, providing work experiences in the private and academic sectors, leading to a recognized qualification (Doctorate).
2. Enable access to first class training in both fundamental and applied research skills to early-stage researchers at world-class academic centres and market leaders in the private sector.
3. Build on the experience gained from previous Copernicus/GMES Research and Development projects in the land monitoring and emergency services areas, including FP5 BIOPRESS, FP6 Geoland and BOSS4GMES, FP7 Geoland2, SAFER and RESPOND.
4. Develop a collaborative training network, through the placement of researchers supported by the grant for short periods (3 months) in other GIONET organisations, and by hosting nationally funded students on request.
5. Provide a flexible training environment with personal career development planning, that early stage researchers can adapt to their cultural, disciplinary and gender needs and in respect of their national education systems.
Since the beginning of the project in 2011, GIONET has accomplished many scientific breakthroughs and technological innovations. In the field of forest monitoring, GIONET produced an algorithm that improved the forest monitoring of the Congo basin using Synthetic Aperture Radar (SAR). James Wheeler at the University of Leicester analysed a time series of 12 ALOS-PALSAR scenes in the Democratic Republic of Congo. He mapped the entire Congo Basin and produced yearly forest cover maps of a five degree by five degree region. He was invited to the 21st Science Team meeting of the Kyoto and Carbon Initiative held in Kyoto, Japan in 2014. Pedro Rodriguez Veiga at the University of Leicester investigated a data mining approach based on the Maximum Entropy algorithm to map forest aboveground biomass in Mexico, Siberia and Spain, with spatial uncertainty and forest probability maps. He demonstrated that optical, SAR and elevation data contain different information that can be combined by this approach to produce better estimations than with a single sensor approach. This method is now being studied in the European Space Agency project GLOBBIOMASS for global application. Martyna Stelmaszczuk-Górska at the University in Jena developed a large-scale forest monitoring scheme and evaluated different biomass retrieval models over Siberian boreal forests. The results are published in the peer-reviewed journal the Siberian Journal of Forest Science.
In the area of land cover and change, Bernard Spies at Airbus DS (formerly Astrium UK) used Random Forest Classifiers for TerraSAR-X (X-band), ENVISAT ASAR (C-band) and ALOS PALSAR (L-band). Two test sites were studied for wet and dry seasons, Cameroon-Dry, Cameroon-Wet, Tanzania-Dry and Tanzania-Wet. The dry season gave higher classification accuracies than the wet season. X-band SAR contributes the most to the classification accuracies of the forested site (Cameroon) and L-band has the highest contribution to the semi-arid (Tanzania) site. The best dual-frequency combination for both sites is L- and X-band. Shailesh Shrestha at IGIK developed a method for using adaptive neighbourhood information for land cover change detection. He used the concept of Geographic Image Object Analysis (GEOBIA) at very local scale in Warsaw, Poland using high-resolution QuickBird images. Multi-scale remote sensing data synergies for land process studies were studied by Maxim Chernetskiy at the University of Jena. His work improved the retrieval of land biophysical and spectral parameter retrieval from remote sensing, data assimilation and radiative transfer modelling.
GIONET advanced the application of remote sensing for coastal zone and freshwater monitoring. Sybrand van Beijma at Airbus DS used airborne radar for salt marsh vegetation mapping in Wales for the first time. His work now forms a reference methodology for the processing of data from the Airbus DS airborne SAR instrument, including the publically-funded AirSAR campaign in the UK in 2014. Dimitris Stratoulias at the Hungarian Academy of Sciences Centre for Ecological Research, Balaton Limnological Institute used airborne hyperspectral data for mapping emergent aquatic plants and studying reed die-back around the lake. The method can be applied to Sentinel-2 satellite data. Stephanie Palmer at the Balaton Limnological Institute explored the use of remote sensing to measure the phytoplankton biomass in the water of Lake Balaton, using a ten year time series of Medium Resolution Imaging Spectrometer (MERIS) satellite imagery, ship-mounted Light Detection and Ranging (LiDAR), and water sampling and laboratory measurements. She demonstrated the feasibility of retrieving accurate and quantitative information on Lake Balaton phytoplankton biomass dynamics through the use of remote sensing techniques.
In the field of geohazards and emergency response, Penelope Kourkouli at GAMMA Remote Sensing developed a hybrid method based on Differential Synthetic Aperture Radar Interferometry (DInSAR) and Persistent Scatterer Interferometry (PSI) and tested it in the Venice Lagoon, Italy and over a semi-arid terrain in Jordan. Jessica Papke at GAMMA investigated spaceborne interferometric Synthetic Aperture Radar (InSAR) data processing and terrestrial radar interferometry for the monitoring of landslide displacements. Topography and surface velocities were mapped for an Alpine study site in the Western Swiss Alps with satellite and ground-based radar interferometry. The surface velocity of an active rock glacier was also mapped with a ground-based radar interferometer. Christoff Fourie at the German Aerospace Center (DLR) developed new methods for sample supervised geographic object based image analysis to support emergency response actions. Two new approaches have been prototyped.
In support of climate adaptation and emergency response, Matthew Ofwono at IGIK applied remote sensing methods for the integration of long-term trend analysis in drought detection and monitoring of drought impacts on crop yield. Soil moisture deficit modelling was implemented in the Decision Support System for Agro-technology Transfer (DSSAT) to estimate wheat yield. This new approach has been demonstrated in incorporating to DSSAT model besides of soil moisture deficit index also fAPAR and LAI from Terra/MODIS satellite what significantly reduced uncertainty in wheat yield prediction. Linda Moser at the German Aerospace Center (DLR) monitored wetlands in semi-arid Africa with optical and SAR data over 15 years (2000-2014). Indicators for water stress and associated land use changes, potentially connected to human crisis situations, were deducted from remote sensing.
Expected final results and their potential impact and use
• Doctoral Training: All Early Stage Researchers have stayed in employment for the planned 36 months in total and are now in various stages of completing their PhD studies. Stephanie Palmer and Sybrand van Beijma have already passed their PhD examinations, Dimitris Stratoulias has submitted his PhD thesis and Matthew Ofwono and Shailesh Shresta have declared their intent to submit it within 3 months. All other ESRs are writing up their thesis aiming for completion by September 2015. Four GIONET summer schools were held to train the researchers and external participants.
• Highly Skilled Researchers: Many GIONET researchers have already secured research employment after the end of their Marie Curie fellowships, for example in the University of Twente (NL), University of Leicester (UK) and the private companies Gamma Remote Sensing (CH) and Geospatial Insight (UK).
• Research from GIONET has been published in 22 highly rated journals, including Remote Sensing of Environment, which is considered the top journal in the field of remote sensing (Impact Factor: 4.77). Other journals include: Applied Geography (Impact Factor 3.08) Remote Sensing (Impact Factor: 2.729) International Journal of Applied Earth Observation and Geoinformation (Impact Factor: 2.809) Engineering Geology for Society and Territory (Impact Factor: 0.721) Earth Science Informatics Impact Factor; 0.694 and Atmospheric and Oceanic Physics (Impact Factor 0.597).
• Networking and Academia-Industry Collaboration: Seven new approved associated partners have joined GIONET: National Observatory of Athens, Greece; University of Stirling, UK; Università degli Studi di Padova, Italy; Danko (Poland), EOS Jena (Germany), Envirosense (Hungary) and Blackbridge (Germany).
• Dissemination and Exploitation: A book contract has been signed with Wiley-Blackwell Publishers to disseminate the GIONET results and 14 book chapters have been submitted and are now undergoing peer-review: Balzter, H. (Ed., submitted): “Earth Observation for Land and Emergency Monitoring - Innovative concepts for environmental monitoring from space”. Wiley, London. Furthermore, GIONET methods and results are implemented and further developed in other projects, including the MaxEnt biomass mapping for GLOBBIOMASS, a reference methodology for the processing of Airbus DS airborne SAR data, and new contract work won by GAMMA.