Overview of Three Ground Radar Experiments for Sensitivity Study over Temperate and Tropical Forests
Albinet, Clément¹; Borderies, Pierre¹; Hamadi, Alia²; Koleck, Thierry³; Floury, Nicolas^4
1ONERA, FRANCE; ²CESBIO, FRANCE; ³CNES, FRANCE; ⁴ESA-ESTEC, NETHERLANDS

1.INTRODUCTION
Tropical forests present the major part of the world forest biomass and their changes in biomass by deforestation and/or by forest regeneration affect strongly the terrestrial carbon budget. To measure with accuracy tropical forest biomass and its temporal change is one of the objectives of the BIOMASS mission [1], a candidate for the European Space Agency 7th Earth Explorer Mission. It is well known that low frequency radars may furnish lots of characteristics of forests, and in particular P-band (435 MHz) is often proposed for biomass estimation [2] and BIOMASS will be the first spaceborne Synthetic Aperture Radar (SAR) operating at P-band. Retrieval of bio-physical parameters of forests with remote sensing is nowadays a challenge. In particular, the biomass of the canopy, soil and branches moisture contents are three parameters of interest. To finely study the mechanisms of radar backscattering at P-band and to link the measurements to biophysical characteristics, sensitivity studies with simulated data are usually performed. In this paper, an original approach of experimental sensitivity studies is presented. Its goal is to better understand the influence of trunks, branches and soil moisture content with several measurements of the same site over time; and the influence of biomass, slope and forest height with measurements of several sites on a short period of time. Three proximity experiments at different scales were conducted: TREESCAT, a system that measures the daily evolution of a cedar radar response; TROPISCAT, a system that measures the daily evolution of a plot of tropical forest [3]; and a campaign to acquire vertical tomograms over several plots of pine forest with various configurations.

2.THE TREESCAT EXPERIMENT
This first proximity experiment consists in the observation of one tree with an emitter-receiver located on the roof of ONERA laboratory in Toulouse, South-West of France. This emitter-receiver is remotely controlled via Ethernet, and then a fine temporal survey of the radiometry and the phase scattered by a tree was performed (Figure 1).
Its goal is to measure the radiometry and temporal coherency evolution of the radar backscattering for the cross polarisation. In addition, weather data like temperature, wind speed, rain rate and soil moisture content were measured synchronously.

3.THE TROPISCAT EXPERIMENT
This experiment has been designed to acquire the polarimetric intensity and the complex coherence in HH, VV and HV, together with a vertical imaging capability (tomography), and in time scales ranging from diurnal, weekly, monthly, up to one year of observation and possibly beyond. The instrument allows an automatic measurement cycle. It is developed with a state-of-the-art instrumentation (a Vector Network Analyzer, P-band antennas, a computer and RF switches boxes) to be assembled and installed on a tower overlooking a tropical forest (Figures 2 and 3). A calibration procedure is performed before each acquisition in order to ensure the quality of the results during all the experiment duration (more than one year).
The Guyaflux tower [4] in Paracou, French Guiana, has been selected to support this experiment. This site, managed by INRA (National Agronomical Research Institute), has the following advantages: the French Guiana forest is one of the world forests with highest biomass density and highest number of tree species, thus constituting one of the most complex and challenging forest ecosystems for the BIOMASS mission; the test site (Paracou area) was overflown by TROPISAR campaign in August 2009 [5]; the interpretation of ground based experiment data will allow in depth understanding of the scattering mechanisms in airborne observations and by extension, in the forthcoming spaceborne BIOMASS observations; the Guyaflux tower (55 m) is used for continuous carbon flux measurements. Studies are foreseen to compare the tower flux measurements with outputs of carbon models constrained by the biomass retrieved from TROPISAR; detailed and comprehensive ground data are collected or recorded continuously for the flux experiment (meteorological data, soil moisture) and for the Paracou forest survey (tree measurements).

4.HIGH RESOLUTION IN-SITU TOMOGRAPHY
A lot of work has been done with high resolution imaging at P-band [6] as well as at L-band [7]. However, it remains a lot of interrogation about vertical disposal of scatterers, even with the polarimetric-interferometric techniques [8]. Airborne tomography has a vertical resolution and shows some promising results [9] but the resolution is poor and doesn't allow to clearly isolate the volume contribution from the soil one. To resolve this uncertainty, an experiment similar to the TROPISCAT one, but with mobile antennas since operating from a bucket truck (Figure 4), allows to image various forest locations of Mende forest (Figure 5), South of France, with very high vertical resolution.
The full polarimetric data was simultaneously acquired in order to process high resolution vertical radar images from which were extracted average vertical profiles and the corresponding integrated values giving the energy respectively backscattered by the soil, volume and the whole scene. Field data like biomass, trees height and density, structure, undergrowth presence and slope were also acquired.

5.ANALYSIS
The three experiments results will be analyzed side by side in order to understand the mechanisms that link the trees moisture contents, the biomass, the forest structure and the other biophysical parameters to the radar backscattering.

6.REFERENCES
[1] BIOMASS Phase 0 Report for Assessment, European Space Agency, Nov. 2008, ref. SP1313/2
[2] Le Toan et al, ''Relating Forest Biomass to SAR data'', IEEETGARS, Volume 30, n°2, pp. 403 - 411, March 1992.
[3] Albinet at al, ''TropiSCAT: A Ground Based Polarimetric Scatterometer Experiment in Tropical Forests'', JSTARS, Volume 5, N° 3, June 2012.
[4] Bonal et al, ''Impact of severe dry season on net ecosystem exchange in the Neotropical rainforest of French Guiana'', Global Change Biology (2008) 14, 1917–1933.
[5] Dubois-Fernandez et al, ''The TropiSAR Airborne Campaign in French Guiana: Objectives, Description, and Observed Temporal Behavior of the Backscatter Signal'', TGRS, August 2012.
[6] Sandberg et al; ''Comparison of L and P band Biomass Retrievals Based on Backscatter from the BIOSAR Campaign'', IGARSS 2009, Cape Town, South-Africa.
[7] Shimada et al, ''PALSAR CALVAL and Generation of the Continent Scale Mosaic Products for Kyoto and Carbon Projects'', IGARSS 2008, Boston, USA.
[8] Garestier et al, '' Polar and PolInSAR analysis of pine forest at L and P band on high resolution data '', IGARSS 2005, Séoul, Korea.
[9] Huang et al, ''Polarimetric SAR Tomography of Tropical Forests at P-Band'', IGARSS 2011, Vancouver, Canada.