Forest Biophysical Parameter Retrieval & Validation

Biophysical parameters like Leaf Area Index (LAI) help scientists to describe and understand ecosystems, and their mass and energy fluxes. This is for example necessary to incorporate them in Global Circulation Models (GCM) as used by the IPCC to establish possible scenarios of future climate change. Their principal advantage lies in their comparability across observation platforms.

Compared to other vegetation types forests introduce specific challenges for the estimation of biophysical parameters through their pronounced 3D structure. Additionally, seasonal variation requires repeated observations to track the development that alter forest properties over short time spans.

This project aims to tackle these challenges by:

  • Exploiting multi-mission satellite observations;
  • Designing a reference site that provides ground-truth with high temporal resolution.

Multi-sensor Biophysical Parameter Retrieval

More and more Earth Observation satellite sensors with decametre resolution (Sentinel-2A & B, Landsat 8) allow forest monitoring on unprecedented scales. However, these missions suffer from poor temporal resolution when considered separately and in respect to dynamic land processes and with cloud cover restrictions. Hence they need be combined – fused – to characterise dynamic processes such as leaf flush and senescence.

For this project we consider fusion methods that make use of several missions:

  • S2A
  • L7 & L8
  • SPOT 6
  • UK-DMC
  • Deimos-1

SPOT, UK-DMC and Deimos-1 data is available for research organisations in the Netherlands through the Dutch Satellite Portal (Satellietdataportaal).

The retrieval methods need to be applicable to all these sensors. Furthermore, they should provide a measure of quality next to the mean value for the estimated parameter to evaluate its reliability with respect to the other sensors. Verrelst et al. (2015) provide a comprehensive overview of retrieval methods and conclude that non-parametric regression methods trained on physical-based radiative transfer models as flexible and fast retrieval methods.

Speulderbos reference site

Ground reference sites are necessary to validate satellite-derived data sets. However, sites that provide biophysical parameters for forests with high revisit time (e.g. weekly) are lacking. The main reason for this are labour intense sampling methods. The aim for the Speulderbos reference site is to test new monitoring approaches that allow constant measurements and to compare them with traditional and recent methodologies.

The following ground-based instruments are available:

  • Digital Hemispherical Photography with Nikon D7000 and Sigma 4.5 mm/F2.8 Fisheye (traditional)
  • Terrestrial Laser Scanner RIEGL VZ-400 (recent)
  • PAI Autonomous System from Transmittance Sensors at 57° (PASTIS57) (monitoring)
  • VEGNET In-situ Monitoring Lidar (monitoring)

Additionally, litter traps are installed as a means to get a direct measure of leaf area. Leaf sampling and analysis with spectrometers was implemented to characterise leaf biochemical traits.

Apart from the ground-based instruments, Unmanned Aerial Vehicles (UAV) are used to produce high resolution imagery. These data sets are used to simulate space-borne sensors. Different sensors are available:

The reference datasets are available for analysis. A quick overview of acquired data can be found here

2016-06-23_data availability.png

Please contact Benjamin Brede for further details.


This work is carried out as part of the IDEAS+ contract funded by ESA-ESRIN.

Special thanks go to Staatsbosbeheer for granting access to the site and to the Wageningen UR Unmanned Aerial Remote Sensing Facility (UARSF) for providing the UAV mapping systems.