Final Report Summary - ADVANCED_SAR (Advanced Techniques for Forest Biomass and Biomass Change Mapping Using Novel Combination of Active Remote Sensing Sensors)
Executive Summary:
The aim of the Advanced_SAR project was to provide high-quality, beyond state-of-the-art techniques for mapping forest canopy height, biomass and biomass change by achieving the best possible cost-efficiency performance out of the given satellite imagery (both SAR and optical), and terrestrial and airborne remote sensing data. The main hypothesis of the project was that remote sensing techniques capable of extracting 3D data can significantly improve the quality of wide coverage forest maps compared with medium-resolution 2D satellite images, such as the Landsat satellite data.
One of the key components in forest biomass estimation is that good quality local ground truth data is needed for training the satellite based biomass estimation models; therefore, in the project, forest test plot inventory techniques based on Terrestrial Laser Scanning systems were studied and developed further to improve the GNSS positioning accuracy of mobile laser scanning systems. In addition, a profiling radar system (FGI-Tomoradar) was designed and a prototype system was build and tested. FGI-Tomoradar is a scientific tool that can be used in the future to possibly better understand the interaction of radar beam with forest structures. The main advancement of the FGI-Tomoradar compared with earlier profiling radars is the integration of accurate Inertial Measurement Unit (IMU) to have better knowledge of the beam position on the ground than in earlier systems.
A comprehensive set of both satellite and airborne remote sensing data were collected, including SAR satellite images (TerraSAR-X, TanDEM-X), very-high-resolution optical satellite images (WorldView, Pleiades), airborne aerial images, and airborne laser scanning data, all capable to produce 3D information, and 3D metrics of the forested test site in the project. The accuracies of various 3D techniques in estimating forest inventory variables were compared and the results have been published in international scientific journals, and have reached good coverage in the scientific community.
During the project, new techniques and results were presented to the end-user community (forest sector, forest inventory authorities) to obtain feed-back of the potential. For example, TanDEM-X wall-to-wall INSAR data of Sweden showed very good potential in forest Above Ground Biomass mapping by providing good biomass estimation accuracy and may increase the update interval of nationwide forest maps, if such a bi-static INSAR data would be available on the operational basis. It would really advantageous for Europe to have bi-static SAR satellite system in Copernicus for mapping changes in forest biomass and in other topographic mapping purposes as well. From the forest sector point of view, Mobile Terrestrial Laser Scanning is very potential technique, by providing with the end-users sufficient results with decreased costs in forest field plot surveys, and will most likely have a breakthrough in forest sector in the near future. The exploitation aspects of the project were discussed and analysed in the Common Exploitation Workshop, in which a commercialization expert appointed by the EC guided project researchers in finding potential exploitable results from the project.
Project Context and Objectives:
The amount of carbon bound in to the forests is one of the key variables in the climate discussion at the global level. The major part of the total forest carbon storage is in the growing stock; therefore, one of the biggest challenges currently is to measure and monitor forest biomass and its changes as accurately and as cost-effectively as possible.
Global forest maps are typically based on the use of optical satellite images (e.g. Landsat-8 and Sentinel-2) as one of the input sources. Optical satellite images can cover large areas operatively with the spatial resolution of some 10 to 20 meters using the visible light region. There are also a lot of research and demonstration results on the use of radar images (microwave) in forest mapping. For example, using Landsat optical images, changes in forest cover can be mapped relatively well. See for example the University of Maryland forest change maps (https://earth-engine¬partners.appspot.com/science-2013-global-forest). However, the main challenge is that optical images are not capable to estimate forest biomass accurate, especially at high biomass volume of forests. This challenge is typically referred to as a saturation problem, where all biomass volume areas higher than some threshold reflectance value appear similar in image data. The same saturation problem applies also to SAR backscatter intensity based biomass estimations (depends on the radar wavelength though). Therefore, to improve the accuracy of biomass estimations, some other prediction metrics are needed. In the past decades, there has been a lot of discussion on the use of alternative remote sensing techniques, which make use of 3D remote sensing metrics to avoid the saturation problem. The 3D metrics from the Airborne Laser Scanning (ALS) has been a breakthrough technique in nationwide forest inventories in many countries.
The main hypothesis of the Advanced_SAR project was that wide-area mapping of forest biomass and its changes can be significantly improved by using 3D capable satellite remote sensing techniques and combining this satellite 3D metrics to cost-efficient new ground survey techniques effectively. In the project, modern satellite data capable of extracting forest height related information are used and compared, and new technologies are developed.
The amount of carbon bound in to the forests is one of the key variables in the climate discussion at the global level. The major part of the total forest carbon storage is in the growing stock; therefore, one of the biggest challenges currently is to measure and monitor forest biomass and its changes as accurately and as cost-effectively as possible.
Global forest maps are typically based on the use of optical satellite images (e.g. Landsat-8 and Sentinel-2) as one of the input sources. Optical satellite images can cover large areas operatively with the spatial resolution of some 10 to 20 meters using the visible light region. There are also a lot of research and demonstration results on the use of radar images (microwave) in forest mapping. For example, using Landsat optical images, changes in forest cover can be mapped relatively well. See for example the University of Maryland forest change maps (https://earth-engine¬partners.appspot.com/science-2013-global-forest). However, the main challenge is that optical images are not capable to estimate forest biomass accurate, especially at high biomass volume of forests. This challenge is typically referred to as a saturation problem, where all biomass volume areas higher than some threshold reflectance value appear similar in image data. The same saturation problem applies also to SAR backscatter intensity based biomass estimations (depends on the radar wavelength though). Therefore, to improve the accuracy of biomass estimations, some other prediction metrics are needed. In the past decades, there has been a lot of discussion on the use of alternative remote sensing techniques, which make use of 3D remote sensing metrics to avoid the saturation problem. The 3D metrics from the Airborne Laser Scanning (ALS) has been a breakthrough technique in nationwide forest inventories in many countries.
The main hypothesis of the Advanced_SAR project was that wide-area mapping of forest biomass and its changes can be significantly improved by using 3D capable satellite remote sensing techniques and combining this satellite 3D metrics to cost-efficient new ground survey techniques effectively. In the project, modern satellite data capable of extracting forest height related information are used and compared, and new technologies are developed. The concept idea of the project is presented in Figure 1.
Project Results:
The main results of the project are presented in the following, roughly in a chronological order:
•First version of the general end-user requirements were drafted for the forest biomass estimation (WP1, D5.1 D1.1 Spring 2014)
•First results of the forest canopy height and biomass maps based on the satellite SAR data, utilizing both radargrammetry (TSX) and interferometry (TDX). At this point, earlier existing ground truth plots were used in the estimation model. (WP2, D2.1 Summer 2014)
•Starting point for the construction of the FGI Tomoradar (light-weight profiling radar for helicopter, or possibly for UAV, use) was designed and plan for data acquisitions. Preliminary design of the system in shown in WP3, D3.1 Spring 2017.
•Construction of the MLS system having SLAM technology started, figure if the FGI-ROAMER in field test below (WP5, D5.2 Autumn 2014)
•Updated field reference collected from super test sites planned and test site data collections (WP7, D7.1 Spring-Autumn 2014)
•Remote sensing data (space-borne, airborne and terrestrial RS data) acquired from super test sites (WP2,3,4,7, D7.1 Summer 2014)
•International comparison of tree extraction using TLS has started. Approval from the EuroSDR as having the status to enlarge research consortium in that area (WP5, Autumn 2014).
•The first result about the Evo 2014 comprehensive data sets is ready, showing that 3D remote sensing is powerful for forest inventory. The comparison was performed at plot level using area-based technology. The world largest and one of the first 3D comparison was published in November 2015. (WP2, WP6, WP7, January 2015)
•Project team presented the first results in international conferences (PIA15+ HRIGI15, ISRSE-36, Dragon3 symposium, IGARSS 2015, and EARSeL 2015) and wrote an invited book chapter discussing 3D remote sensing (WP9, Spring 2015)
•Our first results about the use of optical satellite stereo elevation models in forest inventory attribute prediction are ready and under preparation for publication. Deliverables about optical satellite stereo matching were produced in co-operation by SLU, TU Wien, and FGI. And the comparison of three image matching algorithms has been started (WP6, D6.1+D6.4 Spring 2015)
•The second version of the updated user requirements was prepared by the SLU and the 1st end-user workshop took place on 16 June 2015, Uppsala, Sweden (WP1 and WP9, D1.2 Spring 2015).
•The first working prototype of the FGI-Tomoradar system is ready, and first flights in the Evo area were accomplished two successful campaigns. Initial comparison to other 3D remote sensing data has been carried out (WP3, Autumn 2015)
•The market analysis of the topics included in the project and an initial report of the possible business models was prepared. The FGI has prepared a report about the IPR and exploitation aspects of the project, the open science model being the primary model in exploitation (WP9, D9.10 Autumn 2015)
•To enable forest change detection studies data collection campaign for Summer 2016 was planned (WP2+WP6,+WP7, D7.2 Autumn 2015)
•FGI coordinated an international EuroSDR study on TLS data comparison in forest canopy modelling. The TLS benchmarking was announced in Silvilaser 2013 in Beijing and in EuroSDR Science Committee Meeting autumn 2013 meeting. 19 international groups participated the comparison study. The results will be based for all scientific and R&D development related to using TLS in practical forest inventory. (WP7, Autumn 2015-Spring 2016).
•The development of simulation tool for future spaceborne lidar systems started. The first dataset was simulated from the very high point density airborne laser scanning dataset (WP4, D4.1 Spring 2016)
•SLAM-assisted positioning technology for backpack mobile laser scanning was developed, to help the system positioning inside forests. In co-operation with Centre of Excellence in Laser Scanning funding, new backpack laser scanning systems was developed. (WP5, D5.2+D5.3 Spring 2016).
•The EuroSDR comparison on the use of TLS data in forest plot mapping and derivation of tree parameter is closed (WP7, D7.7 Spring 2016).
•Wall-to-wall demonstration with the Swedish National Forest Inventory was started by planning the acquisition process for TanDEM-X bistatic InSAR data covering large parts of Sweden (WP8, Autumn 2016)
•Project was presented in the INSPIRE2016 Symposium In Barcelona, September 2016 (WP1+ WP9, Autumn 2016)
•Project was presented in the GlobBiomass user workshop (WP9, Spring 2017)
•The FGI/Tomoradar system is ready, and the first publication in MDPI Remote Sensing is open access (WP3, Spring 2017). Chen et al. http://www.mdpi.com/2072-4292/9/1/58
•The evaluation of the potential for spaceborne Lidar for biomass estimation is ready and two papers are being submitted (WP4, Spring 2017)
•Project was presented in the 2nd end-user workshop in Remningstorp, Sweden. (WP9, Spring 2017)
•Optical satellite stereo matching comparison is ready; and some issues with some software components were identified. Two scientific publications has been prepared (WP6, D6.3+D6.6 Spring 2017)
•A user questionnaire to include a wider European perspective was carried out, and results were presented in the IUFRO 125 conference in Freiburg, Germany (WP1, D1.3+D1.5 Spring 2017).
•First results of the wall-to-wall forest biomass demonstration are ready and scientific publication was drafted from the work. (SLU/Henrik Persson) (WP8, D8.1+D8.2 Spring-Summer 2017)
•First results from the INSAR based biomass change mapping are ready and a manuscript of the scientific publication has been prepared from the results. Examples of TanDEM-X elevation change map between 2012-2014 are given (WP7, D7.6 Summer-Autumn 2017)
Potential Impact:
The following expected potential impacts of Advanced_SAR were identified in the beginning of the project:
1.To improve the capability to provide 3D information from forest with space-borne system in order to improve global and regional forest biomass estimates using the canopy height information derived.
2.To develop possible innovative new Copernicus products or applications combining in a novel manner existing and upcoming sensor data and in-situ data.
3.To demonstrate that high-accuracy forest biomass and biomass change maps are achievable by combining various 3D techniques.
4.To substantiate the needs for new observation techniques to be implemented in the next generation of observation satellites (having SAR or Lidar).
5.To demonstrate the capability of biomass mapping through 3D techniques on real events, to our end-users
To our understanding the project successfully replied to its early expectations above in the following ways.
1.Based on our comparison of 3D satellite remote sensing techniques showed good potential in forest inventory variables estimation. Particularly, TanDEM-X INSAR data was found as potential source for nationwide forest mapping (wall-to-wall demo of Sweden). However, there were some challenges with the TDX data, such as elevation offsets and seasonal weather variations in the resulting DEMs, which also have an effect on the output biomass maps. In addition, the continuity of such a bi-static SAR data is not secure, which makes this data less favourable for mapping from the operational forestry point of view. VHR optical stereo was found as a good data source in small-area forest mapping, and was comparable to aerial stereo imagery in terms of biomass estimation accuracy. However, for operational forest activities and, for example, in nationwide mapping, optical data has challenges, such as the problem of clouds and high cost of optical VHR stereo data.
2.We participated in preparation of the ESA Earth Explorer 9 proposal and contributed to the forest mapping part of the consortium, which may lead to new Copernicus services in the long run. However, in November 2017, ESA made a decision not to select this proposal as the EE9 mission.
3.Comprehensive data for comparison studies was collected in three test sites (Evo, Krycklan, Remningstorp), and their results have been published in international scientific journals. One of the first 3D comparisons was given by Yu et al. http://www.mdpi.com/2072-4292/7/12/15809. It should be noted, that results mainly cover boreal forest zone data (pine and spruce dominated stands), and cannot be directly used in other forest types. More research is still needed to better understand the seasonal, weather, forest type, density variations, for example, to the radar penetration into the forest canopy, before more general models can be designed. However, similar results on the use of TDX INSAR data for forest biomass mapping have been obtained by other research groups all around the world.
4.The use of Terrestrial Laser Scanning data and mobile platforms is seen as one of the next potential breakthroughs in the forest inventory sector. Our international EuroSDR comparison is showing the business potential of the developed technology.
5.The development of the profiling radar (FGI-Tomoradar) has gained a lot of international research interest and may lead to better understanding of radar forest response. We however expect that the FGI-Tomoradar has potential as scientific tool in the remote sensing community. The desing and development of the FGI-Tomoradar was given by Chen et al. http://www.mdpi.com/2072-4292/9/1/58
Because the project in general is close to research at the relatively low technological readiness level, it is not very straightforward to estimate socio-economic impacts at this point. For example, TanDEM-X INSAR type of SAR data can be seen as the very potential data source for wide-area/continental scale forest inventory, which could enable improved accuracy with very good temporal resolution (annual change maps). If this kind of bi-static SAR satellite mission could be implemented into the European Copernicus service, there might be business potential in forest inventory sector, not only in Europe, but globally as well. However, it should be noted that business success in this area would also require political support in the form of continuous satellite data from the bistatic INSAR mission. In addition, we can expect that Mobile Terrestrial Laser Scanning systems will have a breakthrough in forest inventories in near future. With mobile systems, sufficient accuracy with lower operational costs can be achieved in forest inventory surveys, which is of high interest in the forest sector. Because business success in terrestrial systems is not dependant on the European-wide political decisions, such as the satellite missions, there is high business value foreseen for such a terrestrial laser scanning systems.
Related to the project topics, two start-up companies have been established during the project:
•Solid Potato Oy (http://solidpotato.com) is a Finnish start-up company and it was founded in February 2015 by a group of world leading scientist in the field of laser scanning. They offer measurement and data processing services as well as hardware are and software solutions for laser scanning applications, where forest surveys if one of the key areas.
•Geoanalysis Sweden AB (http://www.geoanalysis.se) founded in 2017, which however is not only a direct outcome of the Advanced SAR project. Its focus is to provide forest companies and end-users with tools and analyses of their forest, based on remote sensing methods, and to provide other companies SAR expertise to be implemented in their services.
List of Websites:
Project www-site: www.fgi.fi/advancedsar
Videos:
•Evaluation and Derivation of Forest Biomass with Mobile Laser Scanning (https://www.youtube.com/watch?v=h0oEu-bzLn0)
•Presentation at the INSPIRE 2016 conference (https://www.youtube.com/watch?v=Ffea-f3-3sA)
List of beneficiaries, and contact names in technical topics:
Coordinator: FGI, The Finnish Geodetic Institute of the National Land Surveys of Finland (Official name in the EU register: MAANMITTAUSLAITOS , PIC code 964526388) Geodeetinrinne 2, 02431 MASALA, KIRKKONUMMI, FINLAND
Number Partner Organisation Name Acronym
1 MAANMITTAUSLAITOS FGI
Contact Person(s) Telephone Telefax Email Address Responsibility
Juha Hyyppä +358 29 5308026 +358 09 295 55 211 juha.hyyppa@nls.fi (official mail)
juha.coelasr@gmail.com (other mail) The Coordinator of the project including scientific coordination
Mika Karjalainen +358 29 530 8035 +358 09 295 55 211 mika.karjalainen@nls.fi Project manager at the FGI assisting the coordinator
Heli Honkanen +358 29 530 8024 +358 09 295 55 211 heli.honkanen@nls.fi General project management and coordination
SVERIGES LANTBRUKSUNIVERSITET (SLU), Box 7070, UPPSALA, 75007, SWEDEN
Number Partner Organisation Name Acronym
2 SVERIGES LANTBRUKSUNIVERSITET SLU
Contact Person(s) Telephone Telefax Email Address Responsibility
Håkan Olsson +46 90 7868376
+46 70 5535701 Hakan.Olsson@slu.se Point of contact
Johan Fransson +46 90 7868531
+46 70 6608697 Johan.Fransson@slu.se Point of contact
TECHNISCHE UNIVERSITAET WIEN (TU WIEN), Karlsplatz 13, WIEN, 1040, AUSTRIA
Number Partner Organisation Name Acronym
3 TECHNISCHE UNIVERSITAET WIEN TU WIEN
Contact Person(s) Telephone Telefax Email Address Responsibility
Wolfgang Wagner +43 1 58801 12225 wolfgang.wagner@geo.tuwien.ac.at
Markus Hollaus +43 1 58801 12239 Markus.Hollaus@geo.tuwien.ac.at Point of contact
TreeMetrics Ltd, Rubicon Centre, CIT Campus, Bishopstown, CORK, IRELAND
Number Partner Organisation Name Acronym
4 TreeMetrics Ltd TM
Contact Person(s) Telephone Telefax Email Address Responsibility
Enda Keane +353 87 7400786 ekeane@treemetrics.com
Garret Mullooly gmullooly@treemetrics.com
Alex Poveda jpoveda@treemetrics.com Point of contact
CHALMERS TEKNISKA HOEGSKOLA AB, GÖTEBORG, 41296, SWEDEN
Number Partner Organisation Name Acronym
5 CHALMERS TEKNISKA HOEGSKOLA Chalmers
Contact Person(s) Telephone Telefax Email Address Responsibility
Leif Eriksson +46 31 7724856 leif.eriksson@chalmers.se Point of contact
Maciej Soja maciej.soja@chalmers.se Point of contact
Division of work in terms of project topics (Work Package leaders):
WP number WP Name Lead Partner Representative
WP1 User Requirements and User’s Satisfaction SLU Håkan Olsson
WP2 Derivation of features from SAR Chalmers Leif Eriksson
WP3 Physical modelling of the space-borne SAR response TU WIEN Wolfgang Wagner
WP4 Assessing the potential of data from futurespace-borne LiDAR SLU Håkan Olsson
WP5 Development of the automated plotwise field system based on mobile laser scanning FGI Juha Hyyppä
WP6 Derivation of featured from optical satellite imagery TU WIEN Wolfgang Wagner
WP7 Verification of the retrieval methods in super test sites for AGB and AGB change FGI Juha Hyyppä
WP8 Wall-to-wall demonstration with the Swedish National Forest Inventory SLU Håkan Olsson
WP9 Promotion of the results, exploitation preparation FGI Juha Hyyppä
WP10 Project Management FGI Mika Karjalainen
WP11 Scientific and Technical Coordination FGI Juha Hyyppä
The aim of the Advanced_SAR project was to provide high-quality, beyond state-of-the-art techniques for mapping forest canopy height, biomass and biomass change by achieving the best possible cost-efficiency performance out of the given satellite imagery (both SAR and optical), and terrestrial and airborne remote sensing data. The main hypothesis of the project was that remote sensing techniques capable of extracting 3D data can significantly improve the quality of wide coverage forest maps compared with medium-resolution 2D satellite images, such as the Landsat satellite data.
One of the key components in forest biomass estimation is that good quality local ground truth data is needed for training the satellite based biomass estimation models; therefore, in the project, forest test plot inventory techniques based on Terrestrial Laser Scanning systems were studied and developed further to improve the GNSS positioning accuracy of mobile laser scanning systems. In addition, a profiling radar system (FGI-Tomoradar) was designed and a prototype system was build and tested. FGI-Tomoradar is a scientific tool that can be used in the future to possibly better understand the interaction of radar beam with forest structures. The main advancement of the FGI-Tomoradar compared with earlier profiling radars is the integration of accurate Inertial Measurement Unit (IMU) to have better knowledge of the beam position on the ground than in earlier systems.
A comprehensive set of both satellite and airborne remote sensing data were collected, including SAR satellite images (TerraSAR-X, TanDEM-X), very-high-resolution optical satellite images (WorldView, Pleiades), airborne aerial images, and airborne laser scanning data, all capable to produce 3D information, and 3D metrics of the forested test site in the project. The accuracies of various 3D techniques in estimating forest inventory variables were compared and the results have been published in international scientific journals, and have reached good coverage in the scientific community.
During the project, new techniques and results were presented to the end-user community (forest sector, forest inventory authorities) to obtain feed-back of the potential. For example, TanDEM-X wall-to-wall INSAR data of Sweden showed very good potential in forest Above Ground Biomass mapping by providing good biomass estimation accuracy and may increase the update interval of nationwide forest maps, if such a bi-static INSAR data would be available on the operational basis. It would really advantageous for Europe to have bi-static SAR satellite system in Copernicus for mapping changes in forest biomass and in other topographic mapping purposes as well. From the forest sector point of view, Mobile Terrestrial Laser Scanning is very potential technique, by providing with the end-users sufficient results with decreased costs in forest field plot surveys, and will most likely have a breakthrough in forest sector in the near future. The exploitation aspects of the project were discussed and analysed in the Common Exploitation Workshop, in which a commercialization expert appointed by the EC guided project researchers in finding potential exploitable results from the project.
Project Context and Objectives:
The amount of carbon bound in to the forests is one of the key variables in the climate discussion at the global level. The major part of the total forest carbon storage is in the growing stock; therefore, one of the biggest challenges currently is to measure and monitor forest biomass and its changes as accurately and as cost-effectively as possible.
Global forest maps are typically based on the use of optical satellite images (e.g. Landsat-8 and Sentinel-2) as one of the input sources. Optical satellite images can cover large areas operatively with the spatial resolution of some 10 to 20 meters using the visible light region. There are also a lot of research and demonstration results on the use of radar images (microwave) in forest mapping. For example, using Landsat optical images, changes in forest cover can be mapped relatively well. See for example the University of Maryland forest change maps (https://earth-engine¬partners.appspot.com/science-2013-global-forest). However, the main challenge is that optical images are not capable to estimate forest biomass accurate, especially at high biomass volume of forests. This challenge is typically referred to as a saturation problem, where all biomass volume areas higher than some threshold reflectance value appear similar in image data. The same saturation problem applies also to SAR backscatter intensity based biomass estimations (depends on the radar wavelength though). Therefore, to improve the accuracy of biomass estimations, some other prediction metrics are needed. In the past decades, there has been a lot of discussion on the use of alternative remote sensing techniques, which make use of 3D remote sensing metrics to avoid the saturation problem. The 3D metrics from the Airborne Laser Scanning (ALS) has been a breakthrough technique in nationwide forest inventories in many countries.
The main hypothesis of the Advanced_SAR project was that wide-area mapping of forest biomass and its changes can be significantly improved by using 3D capable satellite remote sensing techniques and combining this satellite 3D metrics to cost-efficient new ground survey techniques effectively. In the project, modern satellite data capable of extracting forest height related information are used and compared, and new technologies are developed.
The amount of carbon bound in to the forests is one of the key variables in the climate discussion at the global level. The major part of the total forest carbon storage is in the growing stock; therefore, one of the biggest challenges currently is to measure and monitor forest biomass and its changes as accurately and as cost-effectively as possible.
Global forest maps are typically based on the use of optical satellite images (e.g. Landsat-8 and Sentinel-2) as one of the input sources. Optical satellite images can cover large areas operatively with the spatial resolution of some 10 to 20 meters using the visible light region. There are also a lot of research and demonstration results on the use of radar images (microwave) in forest mapping. For example, using Landsat optical images, changes in forest cover can be mapped relatively well. See for example the University of Maryland forest change maps (https://earth-engine¬partners.appspot.com/science-2013-global-forest). However, the main challenge is that optical images are not capable to estimate forest biomass accurate, especially at high biomass volume of forests. This challenge is typically referred to as a saturation problem, where all biomass volume areas higher than some threshold reflectance value appear similar in image data. The same saturation problem applies also to SAR backscatter intensity based biomass estimations (depends on the radar wavelength though). Therefore, to improve the accuracy of biomass estimations, some other prediction metrics are needed. In the past decades, there has been a lot of discussion on the use of alternative remote sensing techniques, which make use of 3D remote sensing metrics to avoid the saturation problem. The 3D metrics from the Airborne Laser Scanning (ALS) has been a breakthrough technique in nationwide forest inventories in many countries.
The main hypothesis of the Advanced_SAR project was that wide-area mapping of forest biomass and its changes can be significantly improved by using 3D capable satellite remote sensing techniques and combining this satellite 3D metrics to cost-efficient new ground survey techniques effectively. In the project, modern satellite data capable of extracting forest height related information are used and compared, and new technologies are developed. The concept idea of the project is presented in Figure 1.
Project Results:
The main results of the project are presented in the following, roughly in a chronological order:
•First version of the general end-user requirements were drafted for the forest biomass estimation (WP1, D5.1 D1.1 Spring 2014)
•First results of the forest canopy height and biomass maps based on the satellite SAR data, utilizing both radargrammetry (TSX) and interferometry (TDX). At this point, earlier existing ground truth plots were used in the estimation model. (WP2, D2.1 Summer 2014)
•Starting point for the construction of the FGI Tomoradar (light-weight profiling radar for helicopter, or possibly for UAV, use) was designed and plan for data acquisitions. Preliminary design of the system in shown in WP3, D3.1 Spring 2017.
•Construction of the MLS system having SLAM technology started, figure if the FGI-ROAMER in field test below (WP5, D5.2 Autumn 2014)
•Updated field reference collected from super test sites planned and test site data collections (WP7, D7.1 Spring-Autumn 2014)
•Remote sensing data (space-borne, airborne and terrestrial RS data) acquired from super test sites (WP2,3,4,7, D7.1 Summer 2014)
•International comparison of tree extraction using TLS has started. Approval from the EuroSDR as having the status to enlarge research consortium in that area (WP5, Autumn 2014).
•The first result about the Evo 2014 comprehensive data sets is ready, showing that 3D remote sensing is powerful for forest inventory. The comparison was performed at plot level using area-based technology. The world largest and one of the first 3D comparison was published in November 2015. (WP2, WP6, WP7, January 2015)
•Project team presented the first results in international conferences (PIA15+ HRIGI15, ISRSE-36, Dragon3 symposium, IGARSS 2015, and EARSeL 2015) and wrote an invited book chapter discussing 3D remote sensing (WP9, Spring 2015)
•Our first results about the use of optical satellite stereo elevation models in forest inventory attribute prediction are ready and under preparation for publication. Deliverables about optical satellite stereo matching were produced in co-operation by SLU, TU Wien, and FGI. And the comparison of three image matching algorithms has been started (WP6, D6.1+D6.4 Spring 2015)
•The second version of the updated user requirements was prepared by the SLU and the 1st end-user workshop took place on 16 June 2015, Uppsala, Sweden (WP1 and WP9, D1.2 Spring 2015).
•The first working prototype of the FGI-Tomoradar system is ready, and first flights in the Evo area were accomplished two successful campaigns. Initial comparison to other 3D remote sensing data has been carried out (WP3, Autumn 2015)
•The market analysis of the topics included in the project and an initial report of the possible business models was prepared. The FGI has prepared a report about the IPR and exploitation aspects of the project, the open science model being the primary model in exploitation (WP9, D9.10 Autumn 2015)
•To enable forest change detection studies data collection campaign for Summer 2016 was planned (WP2+WP6,+WP7, D7.2 Autumn 2015)
•FGI coordinated an international EuroSDR study on TLS data comparison in forest canopy modelling. The TLS benchmarking was announced in Silvilaser 2013 in Beijing and in EuroSDR Science Committee Meeting autumn 2013 meeting. 19 international groups participated the comparison study. The results will be based for all scientific and R&D development related to using TLS in practical forest inventory. (WP7, Autumn 2015-Spring 2016).
•The development of simulation tool for future spaceborne lidar systems started. The first dataset was simulated from the very high point density airborne laser scanning dataset (WP4, D4.1 Spring 2016)
•SLAM-assisted positioning technology for backpack mobile laser scanning was developed, to help the system positioning inside forests. In co-operation with Centre of Excellence in Laser Scanning funding, new backpack laser scanning systems was developed. (WP5, D5.2+D5.3 Spring 2016).
•The EuroSDR comparison on the use of TLS data in forest plot mapping and derivation of tree parameter is closed (WP7, D7.7 Spring 2016).
•Wall-to-wall demonstration with the Swedish National Forest Inventory was started by planning the acquisition process for TanDEM-X bistatic InSAR data covering large parts of Sweden (WP8, Autumn 2016)
•Project was presented in the INSPIRE2016 Symposium In Barcelona, September 2016 (WP1+ WP9, Autumn 2016)
•Project was presented in the GlobBiomass user workshop (WP9, Spring 2017)
•The FGI/Tomoradar system is ready, and the first publication in MDPI Remote Sensing is open access (WP3, Spring 2017). Chen et al. http://www.mdpi.com/2072-4292/9/1/58
•The evaluation of the potential for spaceborne Lidar for biomass estimation is ready and two papers are being submitted (WP4, Spring 2017)
•Project was presented in the 2nd end-user workshop in Remningstorp, Sweden. (WP9, Spring 2017)
•Optical satellite stereo matching comparison is ready; and some issues with some software components were identified. Two scientific publications has been prepared (WP6, D6.3+D6.6 Spring 2017)
•A user questionnaire to include a wider European perspective was carried out, and results were presented in the IUFRO 125 conference in Freiburg, Germany (WP1, D1.3+D1.5 Spring 2017).
•First results of the wall-to-wall forest biomass demonstration are ready and scientific publication was drafted from the work. (SLU/Henrik Persson) (WP8, D8.1+D8.2 Spring-Summer 2017)
•First results from the INSAR based biomass change mapping are ready and a manuscript of the scientific publication has been prepared from the results. Examples of TanDEM-X elevation change map between 2012-2014 are given (WP7, D7.6 Summer-Autumn 2017)
Potential Impact:
The following expected potential impacts of Advanced_SAR were identified in the beginning of the project:
1.To improve the capability to provide 3D information from forest with space-borne system in order to improve global and regional forest biomass estimates using the canopy height information derived.
2.To develop possible innovative new Copernicus products or applications combining in a novel manner existing and upcoming sensor data and in-situ data.
3.To demonstrate that high-accuracy forest biomass and biomass change maps are achievable by combining various 3D techniques.
4.To substantiate the needs for new observation techniques to be implemented in the next generation of observation satellites (having SAR or Lidar).
5.To demonstrate the capability of biomass mapping through 3D techniques on real events, to our end-users
To our understanding the project successfully replied to its early expectations above in the following ways.
1.Based on our comparison of 3D satellite remote sensing techniques showed good potential in forest inventory variables estimation. Particularly, TanDEM-X INSAR data was found as potential source for nationwide forest mapping (wall-to-wall demo of Sweden). However, there were some challenges with the TDX data, such as elevation offsets and seasonal weather variations in the resulting DEMs, which also have an effect on the output biomass maps. In addition, the continuity of such a bi-static SAR data is not secure, which makes this data less favourable for mapping from the operational forestry point of view. VHR optical stereo was found as a good data source in small-area forest mapping, and was comparable to aerial stereo imagery in terms of biomass estimation accuracy. However, for operational forest activities and, for example, in nationwide mapping, optical data has challenges, such as the problem of clouds and high cost of optical VHR stereo data.
2.We participated in preparation of the ESA Earth Explorer 9 proposal and contributed to the forest mapping part of the consortium, which may lead to new Copernicus services in the long run. However, in November 2017, ESA made a decision not to select this proposal as the EE9 mission.
3.Comprehensive data for comparison studies was collected in three test sites (Evo, Krycklan, Remningstorp), and their results have been published in international scientific journals. One of the first 3D comparisons was given by Yu et al. http://www.mdpi.com/2072-4292/7/12/15809. It should be noted, that results mainly cover boreal forest zone data (pine and spruce dominated stands), and cannot be directly used in other forest types. More research is still needed to better understand the seasonal, weather, forest type, density variations, for example, to the radar penetration into the forest canopy, before more general models can be designed. However, similar results on the use of TDX INSAR data for forest biomass mapping have been obtained by other research groups all around the world.
4.The use of Terrestrial Laser Scanning data and mobile platforms is seen as one of the next potential breakthroughs in the forest inventory sector. Our international EuroSDR comparison is showing the business potential of the developed technology.
5.The development of the profiling radar (FGI-Tomoradar) has gained a lot of international research interest and may lead to better understanding of radar forest response. We however expect that the FGI-Tomoradar has potential as scientific tool in the remote sensing community. The desing and development of the FGI-Tomoradar was given by Chen et al. http://www.mdpi.com/2072-4292/9/1/58
Because the project in general is close to research at the relatively low technological readiness level, it is not very straightforward to estimate socio-economic impacts at this point. For example, TanDEM-X INSAR type of SAR data can be seen as the very potential data source for wide-area/continental scale forest inventory, which could enable improved accuracy with very good temporal resolution (annual change maps). If this kind of bi-static SAR satellite mission could be implemented into the European Copernicus service, there might be business potential in forest inventory sector, not only in Europe, but globally as well. However, it should be noted that business success in this area would also require political support in the form of continuous satellite data from the bistatic INSAR mission. In addition, we can expect that Mobile Terrestrial Laser Scanning systems will have a breakthrough in forest inventories in near future. With mobile systems, sufficient accuracy with lower operational costs can be achieved in forest inventory surveys, which is of high interest in the forest sector. Because business success in terrestrial systems is not dependant on the European-wide political decisions, such as the satellite missions, there is high business value foreseen for such a terrestrial laser scanning systems.
Related to the project topics, two start-up companies have been established during the project:
•Solid Potato Oy (http://solidpotato.com) is a Finnish start-up company and it was founded in February 2015 by a group of world leading scientist in the field of laser scanning. They offer measurement and data processing services as well as hardware are and software solutions for laser scanning applications, where forest surveys if one of the key areas.
•Geoanalysis Sweden AB (http://www.geoanalysis.se) founded in 2017, which however is not only a direct outcome of the Advanced SAR project. Its focus is to provide forest companies and end-users with tools and analyses of their forest, based on remote sensing methods, and to provide other companies SAR expertise to be implemented in their services.
List of Websites:
Project www-site: www.fgi.fi/advancedsar
Videos:
•Evaluation and Derivation of Forest Biomass with Mobile Laser Scanning (https://www.youtube.com/watch?v=h0oEu-bzLn0)
•Presentation at the INSPIRE 2016 conference (https://www.youtube.com/watch?v=Ffea-f3-3sA)
List of beneficiaries, and contact names in technical topics:
Coordinator: FGI, The Finnish Geodetic Institute of the National Land Surveys of Finland (Official name in the EU register: MAANMITTAUSLAITOS , PIC code 964526388) Geodeetinrinne 2, 02431 MASALA, KIRKKONUMMI, FINLAND
Number Partner Organisation Name Acronym
1 MAANMITTAUSLAITOS FGI
Contact Person(s) Telephone Telefax Email Address Responsibility
Juha Hyyppä +358 29 5308026 +358 09 295 55 211 juha.hyyppa@nls.fi (official mail)
juha.coelasr@gmail.com (other mail) The Coordinator of the project including scientific coordination
Mika Karjalainen +358 29 530 8035 +358 09 295 55 211 mika.karjalainen@nls.fi Project manager at the FGI assisting the coordinator
Heli Honkanen +358 29 530 8024 +358 09 295 55 211 heli.honkanen@nls.fi General project management and coordination
SVERIGES LANTBRUKSUNIVERSITET (SLU), Box 7070, UPPSALA, 75007, SWEDEN
Number Partner Organisation Name Acronym
2 SVERIGES LANTBRUKSUNIVERSITET SLU
Contact Person(s) Telephone Telefax Email Address Responsibility
Håkan Olsson +46 90 7868376
+46 70 5535701 Hakan.Olsson@slu.se Point of contact
Johan Fransson +46 90 7868531
+46 70 6608697 Johan.Fransson@slu.se Point of contact
TECHNISCHE UNIVERSITAET WIEN (TU WIEN), Karlsplatz 13, WIEN, 1040, AUSTRIA
Number Partner Organisation Name Acronym
3 TECHNISCHE UNIVERSITAET WIEN TU WIEN
Contact Person(s) Telephone Telefax Email Address Responsibility
Wolfgang Wagner +43 1 58801 12225 wolfgang.wagner@geo.tuwien.ac.at
Markus Hollaus +43 1 58801 12239 Markus.Hollaus@geo.tuwien.ac.at Point of contact
TreeMetrics Ltd, Rubicon Centre, CIT Campus, Bishopstown, CORK, IRELAND
Number Partner Organisation Name Acronym
4 TreeMetrics Ltd TM
Contact Person(s) Telephone Telefax Email Address Responsibility
Enda Keane +353 87 7400786 ekeane@treemetrics.com
Garret Mullooly gmullooly@treemetrics.com
Alex Poveda jpoveda@treemetrics.com Point of contact
CHALMERS TEKNISKA HOEGSKOLA AB, GÖTEBORG, 41296, SWEDEN
Number Partner Organisation Name Acronym
5 CHALMERS TEKNISKA HOEGSKOLA Chalmers
Contact Person(s) Telephone Telefax Email Address Responsibility
Leif Eriksson +46 31 7724856 leif.eriksson@chalmers.se Point of contact
Maciej Soja maciej.soja@chalmers.se Point of contact
Division of work in terms of project topics (Work Package leaders):
WP number WP Name Lead Partner Representative
WP1 User Requirements and User’s Satisfaction SLU Håkan Olsson
WP2 Derivation of features from SAR Chalmers Leif Eriksson
WP3 Physical modelling of the space-borne SAR response TU WIEN Wolfgang Wagner
WP4 Assessing the potential of data from futurespace-borne LiDAR SLU Håkan Olsson
WP5 Development of the automated plotwise field system based on mobile laser scanning FGI Juha Hyyppä
WP6 Derivation of featured from optical satellite imagery TU WIEN Wolfgang Wagner
WP7 Verification of the retrieval methods in super test sites for AGB and AGB change FGI Juha Hyyppä
WP8 Wall-to-wall demonstration with the Swedish National Forest Inventory SLU Håkan Olsson
WP9 Promotion of the results, exploitation preparation FGI Juha Hyyppä
WP10 Project Management FGI Mika Karjalainen
WP11 Scientific and Technical Coordination FGI Juha Hyyppä