earth online

New campaign dataset for tomographic sensing experiment

22 May 2024

ESA conducts airborne and ground-based campaigns to support the development of new instruments, and calibration and validation of existing instruments. These campaigns simulate satellite-based instruments and are conducted all over the world in support of a wide range of applications.

The technical assistance for airborne measurements during the tomographic sensing experiment (TomoSense) campaign conducted tomographic and fully polarimetric SAR (Synthetic Aperture Radar) surveys of temperate forests. The goal was to acquire P-, L-, and C-band SAR data to study the features of radar scattering from forests. The use of SAR tomography has presented new opportunities for studying forests from space, and the upcoming Biomass mission is an example, which is expected to perform SAR tomography with a P-band SAR instrument. 

Data being collected at the test site
Data being collected at the test site

The TomoSense goals were to:

  • Assess the potential and quantify the information content of C-band single-pass interferometry and tomography for forest mapping and parameter retrieval Study and record complementarity and synergy of P-, L-, and C-band SAR data
  • Provide initial recommendations for future P-, L-, and C-band missions and concepts
  • Create a unique database for analytical use when performing studies of future Earth observation satellite missions

TomoSense took place in the Kermester area of the Eifel National Park in Germany, between 2019 and 2021. Originally planned to take place during 2019, most of the campaign was delayed to 2020 and 2021, due to unexpected weather conditions and the COVID-19 pandemic. Initial in-situ studies took place in spring 2019, followed by a number of airborne surveys in summer and autumn 2020. Final activities for the campaign took place a year later in autumn 2021.

Despite the challenges posed by this unprecedented situation, the campaign team successfully acquired P-, L-, and C-band SAR data, which was complemented by ground-based observations.

P-band data were acquired during the BelSAR-P campaign in 2020. The L- and C-band surveys took place in September and November 2020, and one final survey in October 2021. 

For each survey, two aircraft were flown simultaneously, acquiring mono- and bi-static SAR data in their respective radar bands, in order to simulate a two-satellite bi-static mission. Fight safety regulations made it impossible to fly the aircraft as closely as a two-satellite constellation would operate, so the along-track baseline expected for the acquisitions was larger than planned. The campaign team considered and proposed alternatives that could be implemented in future airborne campaigns to address this limitation.

The TomoSense campaign provided extensive information about P-, L-, and C-band SAR data of forests. 

C-band conclusions

TomoSense data indicated that C-band radar waves are capable of penetrating down to the ground level. This supports the feasibility of C-band tomography of temperate forests, on the condition that acquisitions are at a temporal baseline of a few tens of milliseconds.

Implementation within a two-satellite bi-static mission would require the satellites to fly in formation with an along-track baseline systematically within a few tens of metres. An existing example of this is DLR’s TanDEM-X mission. However, these requirements may be too demanding for small satellite or opportunity missions. In this case, zero-temporal baseline could only be achieved by using formations of three or more satellites.

P- and L-band conclusions

Both P- and L-band observations indicated sensitivity to the whole vegetation layer, in that both frequencies allow for a clear detection of terrain and forest canopies. Both frequencies are also robust with regards to temporal decorrelation over a few hours, resulting in the possibility to produce high-quality tomographic imaging from repeat-pass campaign data.

Accordingly, no fundamental difference was observed between P- and L-band at the Kermester test site, instead showing different sensitivities to different forest elements. L-band was more sensitive to canopy scattering and P-band to scattering from the terrain level, which includes double-bounce scattering.

Both frequencies demonstrated the fundamental role played by tomographic layering, resulting in a much clearer connection to Above Ground Biomass (AGB) than using intensity data only. The use of normalised tomographic indicators such as fractional volume intensity showed sensitivity to AGB, too. Though only a preliminary result, it indicates a potential to retrieve AGB without absolute radiometric calibration. Further research and testing is required to explore this further but, if feasible, it would make it possible to accurately observe forested areas using formations of small satellites using fine synchronisation during post-processing rather than absolute radiometric calibration of the satellite’s instruments.

Dataset

TomoSense successfully acquired 1800 SAR images, which are processed in Matlab and NetCDF formats. These are available to download in the following formats, to suit different users:

  • Calibrated Single Look Complex SAR images, which are finely coregistered, phase calibrated, and ground steered, to enable researchers to directly implement interferometric or tomographic processing 
  • Tomographic cubes representing forest scattering in 3D with radar and geographical coordinates, useful for non-radar experts 

The dataset also includes Airborne Lidar Scanning-derived maps of forest height and AGB, the forest census, and Terrestrial Laser Scanning profiles.

The TomoSense dataset is expected to provide valuable information for studies and researchers of SAR tomography over forests, and support future campaigns to study different tomographic features in more detail.  

Learn more about the TomoSense campaign and how to request the data

Tweet