Science Service System

Summary of Proposal LAN0191

TitleTERRA-SIBESSC (TerraSAR-X for SIBERIA Earth System Science Cluster)
Investigator Schmullius, Christiane - Friedrich-Schiller University Jena, Department of Geography
Team Members
Prof. Dr. Schmullius, Christiane - Friedrich-Schiller-University Jena, Institute of Geography, Department for Geoinformatics and Remote Sensing
Dr. Voigt, Stefan - Deutsches Zentrum für Luft und Raumfahrt, Deutsches Fernerkundungsdatenzentrum
Prof. Dr. Le Toan, Thuy - Centre d`Etudes Spatiales de la Biosphere, Universite Paul Sabatier
Prof. Dr. Quegan, Shaun - University of Sheffield, Sheffield Center for Earth Observation Science
Dr. Balzter, Heiko - University of Leicester, Department of Geography
Prof. Dr. Wagner, Wolfgang - University of Vienna, IPF Institute of Photogrammetry and Remote Sensing
Dr. Lucht, Wolfgang - Potsdam Institute for Climate Impact Research, Potsdam Institute for Climate Impact Research
Dipl.-Geogr. Gerlach, Roman - FSU, Institute of Geography, Department for Geoinformatics and Remote Sensing
Dr. Santoro, M. - Gamma Remote Sensing Research and Consulting AG,
SummaryTERRA-SIBESSC is proposed in the context of a data and model cluster which is under development using existing results from two large EU-projects, SIBERIA and SIBERIA-II. This cluster is called SIB-ESS-C (SIBERIA Earth System Science Cluster). The objectives of SIB-ESS-C are to create a comprehensive geo-database with EO data, EO data products and GIS layers for the Siberian region and to test and validate methods to derive important EO-products from EO data sets for users on demand. Additionally earth system modelling approaches are integrated into the database interface using modules of established biosphere models. Modelling approaches and Earth observation data products can be used together via a web interface that allows direct access to these models and EO data processing algorithms. TERRA-SIBESSC for SIB-ESS-C is a joint research project between the SIBERIA-II (FP5) project team members coordinated and lead by the Friedrich-Schiller University Jena. The region under study are two large administrational regions in Northern Eurasia, i.e. Krasnoyarsk Kray and Irkutsk Oblast, covering an area of 328 million ha and representing a significant part of the Earth's boreal biome which plays a critical role in global climate. SIB-ESS-C will be an essential contribution to carbon cycle investigations by providing crucial and unique data on vegetation biomass for the Siberian region. TerraSAR-X data will play an important role to test and validate the application of SAR data for the development of Earth observation products. The demand for up to date information about the state of the carbon-cycle will be answered with TERRA-SIBESSC for SIB-ESS-C on a regional scale in a format that is unique. The potential of high resolution TerraSAR-X data for the following information types and products will be analysed for the following purposes: 1. high resolution polarimetric X-band analysis for forest cover change, small-scale disturbance mapping and clear cut mapping, 2. texture analysis for tree type identification and forest type mapping, 3. large scale forest cover mapping, and 4. attempt for biomass estimation based on multitemporal and volume scattering effects.
Detailed reportIn the following, the general project structure will be described to receive an overview of the involved projects. At this stage, TerraSAR-X data play an important role to test and validate the application of high resolution SAR data for the development of unique Earth observation products (biomass, forest cover and forest cover change, disturbances). The data is used within the ESA project GSE Forest Monitoring, Detection and Post-Monitoring of Natural and Human Induced Forest Disturbances, Forest Monitoring in Russia. GSE FM is one element of the GMES (Global Monitoring for Environmental and Security) initiative of the ESA Earthwatch Programmes. Main goal is to deliver customised and policy-relevant information mainly based on EO data in ready-to-use packages in the field of Climate Change, Sustainable Forest Management as well as in Environmental Issues and Natural Protection. The supplied products and services are validated and standardised to support decision-making and improved policies that enable cost effective sustainable forest management in various countries. To guarantee the GSE FM standards all products including their specifications and instructions for production are collected within the Service Portfolio Specifications. The GSE FM consortium contains 18 Service Provider and 26 End-User and is lead by GAF AG, Germany. Each Service Provider is responsible for a so called “Service Case”. In an official agreement (SLA-Service Level Agreement) a detailed description of the service and the products for the specific End-User is formalized with the Service Provider. In case of the service for Russia, the service specification is described in the following. The Irkutsk Oblast is dominated by typical taiga forest. 82% of the territory is covered by forest, which corresponds to an area of 71.5 million hectares or 9.9% of Russian forested areas. Within the Irkutsk Oblast the State Forest Agency of Irkutsk executes the policy for the Ministry of Natural Resources of Russian Federation (MNR). Duties of the FS-I include forest use, forest monitoring and control activities, as well as protection and reforestation throughout the Irkutsk region. Forest information are provided by the conduction of periodical forest inventories and intermediate forest enterprises in order to update forest inventory data with respect to changes caused by forest management activities and natural disturbances. According to formal requirements forest inventories must be conducted every 10 -15 years. For updating the SFA (State Forest Account) as well as for management and monitoring purposes an annual update cycle is obligatory. Due to common forest fire events as well as intensive human activities such as clear cutting and cultivation the Irkutsk territory is characterized by intensive large area changes of forests. A frequent full inventory of Irkutsk forest resources is a great challenge. The conduction of ground inventories within this region is rather complicated due to the large size of the region, insufficient infrastructure, inaccessibility of many areas as well as severe climatic conditions. Moreover, ground based forest inventory is costly and time consuming. This GSE FM service provides a powerful tool for effective forest monitoring and inventory at regional scale. Reliable and up-to-date information on forest extent and changes therein will be generated using high-resolution ASAR and PALSAR data. Cross-polarized C-Band SAR data acquired at large incidence angles, cross-polarised L-Band SAR data and estimated coherence data from PALSAR provide a powerful tool for forest mapping issues. This GSE FM service aims to support the FA of GSNR in solving its duties in a more efficient way. The provided service products could be used as part of the regular forest inventory and annual procedure of the SFA as well as to update the existing databases of the FA of GSNR. Within the project three products are produced: 1) Forest Area Map, 2) Forest Area Change Map, 3) Clear Cut and Burned Area Map. These products will be implemented into TERRA-SIBESSC and further into SIB-ESS-C (Siberian Earth System Science Cluster) which is currently being established at the University of Jena (Germany). The objectives of the SIB-ESS-C project are to develop a spatial data infrastructure to facilitate Earth system science studies in Siberia, to set up a web interface to provide access to data products created during the SIBERIA-II project, to continue remote sensing data acquisition and product generation to build up time series for a larger region in Eurasia and to integrate additional products from other projects as well as from external collaborators. The final stage of SIB-ESS-C will provide online geo-visualization and analysis tools (including a biosphere/Earth system modelling interface) for integrated data analysis. After receiving the 25 High Resolution Spot Light TerraSAR-X Data, the data has been geocoded using Gamma Remote Sensing Software. For geocoding, the 100m resolution SRTM has been used. However, this also brings mistakes with itself. The resolution of SRTM is not sufficient in order to find enough ground control points to geocode the SLC data. No offsets are found. Therefore we only worked with the orbit parameter from TerraSAR, which are within an error of 4m in azimuth according to Schubert et al. 2008. The geospatial resolution contains 5 m x 5 m as a multilooking had to be performed in order to reduce speckle. Afterwards, the data were used to perform an accuracy assessment on the produced maps within GSE FM. This Forest Area Map has been produced for the reference year 2007 based on EO ALOS PALSAR cross-polarized L-Band SAR Intensities, HH/HV summer data in the time frame 29.06.2007 to 05.09.2007 and EO ALOS PALSAR co-polarized HH L-Band SAR coherence winter pairs in the time frame 27.12.2006 to 26.02. Herein the classes Forest and Non-Forest Areas as well as a class Sparse Forest (including bushes, grassland, or individual trees) were mapped. The product is specified by a Minimum Mapping Unit of 1 ha. For accuracy assessment 25 high resolution Spotlight Mode TerraSAR-X data were randomly spread over the entire region. For each class, a minimum of 5 sample points per scene were spread over the classification result that is covered by the TerraSAR-X scene using the stratified random method. After spreading the points, the TerraSAR-X scene was overlayed to the points and manually filled with the class values. This could be performed by expert knowledge. Afterwards, the accuracy assessment report was written and an error matrix was conducted. To retrieve the overall accuracy for each class, all 25 assessment reports were transformed into one error matrix. The total Forest Area Map accuracy is 95.13%. As one scene was ordered in different incidence angles, one in 51° and one in 29°, it was possible to analyse the contrast between both angles. As a larger incidence angle is more suitable to differenciate between forest and non-forest (due to shadow effects, visible edges etc.) it is recommended to use large incidence angles for forestry application. Future prospectus using TerraSAR-X Data will comprise an update of the forest inventory database initiated by IIASA. This database is already about 11-15 years old and a solution needs to be found in order to control and update theses test sites. It is foreseen that a SAR timeseries will be acquired covering the annual phenological growth cycle. This will be a sustainable aim of TERRA-SIBESSC. The current question needs to be clarified which TerraSAR-X mode would be best suited for this investigation.

Back to list of proposals

© DLR 2004-2016