GNSS Reflectometry for Soil Moisture and Above Ground Biomass Monitoring
Egido, Alejandro1; Caparrini, Marco1; Paloscia, Simonetta2; Santi, Emanuele2; Guerriero, Leila3; Pierdicca, Nazzareno4; Fontanelli, Giacomo2; Floury, Nicolas5
1Starlab Barcelona, SPAIN; 2CNR-IFAC, Florence, ITALY; 3DISP, Tor Vergata University, Rome, ITALY; 4DIET, La Sapienza University, Rome, ITALY; 5ESA/ESTEC, NETHERLANDS

Several active and passive remote sensing techniques, such as monostatic radars and microwave radiometers, have been proposed to measure key land parameters like soil moisture and vegetation biomass over wide areas. As it is well known, soil moisture is a key parameter in the surface hydrological cycle, which is essential for the understanding of the interaction between land surfaces and the atmosphere. Vegetation biomass plays a crucial role in the carbon cycle through the processes of carbon uptake and respiration. It is therefore a variable of paramount importance for global climate modeling and greenhouse emission inventories. Despite the recognized importance of these parameters, providing the resolution and precision required by most climatological model remains a challenge.

Global Navigation Satellite System Reflectometry (GNSS-R) could represent a suitable solution to this problem since the use of GNSS signals as sources of opportunity allows bistatic radar measurements, which theory and previous experiments showed to be sensitive to land bio-geophysical parameters. The GRASS project (GNSS Reflectometry Analysis for BiomaSS Monitoring) is an activity promoted and funded by the European Space Agency (ESA) aimed to address the feasibility of monitoring vegetation biomass with GNSS-R. The project is divided in two main parts; the first one entails the development of a bistatic scattering model for vegetated areas and its integration within a GNSS-R simulator, which will allow the analysis of the physical mechanisms involved in bistatic scattering. The second one, on which this paper is focused, includes the performance of a set of airborne experimental campaigns with a GNSS-R receiver specially tailored for land applications. During those flights, a wide range of land types with different vegetation covers were observed, which allowed to perform a sensitivity analysis of GNSS reflected signals towards soil moisture and vegetation biomass, and to validate the scattering models also developed in this activity.

The GRASS Experimental Campaigns were carried out over two main observation areas in the vicinity of Florence, Italy; the first one is an agricultural area along the Elsa River, (43.665N, 10.93E), covered with different crop types. The second observation zone, (43.607N, 10.671E) is an area scattered with poplar plots and natural forested areas with high biomass content. The flight campaigns were performed during the first week of July and second week of November 2011 (Day of Year 184-186, and 312-314). This allowed to observe crops and poplar plots in different development stages and with different soil moisture content, enabled by seasonal meteorological conditions in the area of observation. The GNSS-R instrument used for the GRASS experimental campaign was designed and developed at Starlab Barcelona during the ESA project SAM, and upgraded during the GRASS project. The SAM instrument features an up-looking GPS L1 right hand circular polarized (RHCP) antenna for the reception of the direct signal, and two down-looking antennae: a left hand circular polarized (LHCP) and a RHCP in order to perform polarimetric measurements of the GNSS reflected signals. The flight campaigns were complemented with extensive in situ campaigns for the retrieval of ground-truth data, for both vegetation and soil parameters.

The GNSS-R data were processed and the Rrl and Rrr reflectivity components were obtained and geo-referenced based on the position of the aircraft at the moment of the acquisition and the GPS satellite position. The observed variations in the measured reflectivity coefficients in both rl and rr polarizations could be initially linked to features on the surface, such as water bodies, roads, housing, and the presence of vegetation. Significant variations in both Rrl and Rrr reflectivity components could also be related to soil moisture content and above ground biomass.

The Rrl and Rrr polarization ratio showed significant sensitivity and high correlation with respect to soil moisture content with high stability with respect to soil surface roughness. This observable showed a 3 dB variation for a soil moisture range between 5% and 35% with a correlation coefficient of 0.85. In addition, the Rrl reflection coefficient experiences a monotonic decrease up to an above ground biomass of more than 350 t/ha. A sensitivity of 1.5 dB / (100 t/ha) was measured for the observed biomass range, in agreement with scattering models. The fact that GNSS-R observables do not exhibit a sensitivity saturation with increasing biomass constitutes a remarkable improvement with respect to conventional biomass remote sensing techniques, such as monostatic radars, which present a saturation around 100 t/ha. These results suggest the good prospects of GNSS-R for land monitoring applications, and open the door for further developments in this area.