Soil moisture retrieval from multi-incidence angle observations at L-band

Abstract

Soil moisture is an important environmental variable in regulating energy fluxes and water infiltration near the soil surface. This makes it a significant parameter in meteorological and climate modelling applications. While ground measurement of soil moisture at a large spatial scale is cumbersome and time-consuming, remote sensing offers the advantage of frequent observations in time and space. The most promising remote sensing technique for soil moisture observations is microwave radiometry at L-band, which is highly sensitive to moisture and less affected by roughness and canopy attenuation than compared to shorter wavelengths in the microwave range. Consequently, the first dedicated mission for global Soil Moisture and Ocean Salinity mapping (SMOS), launched in November 2009, has a passive microwave radiometer at L-band (1.4 GHz) that uses a two-dimensional interferometric Y-shaped antenna. The novel design of this satellite, yielding multi-incidence angle observations, provides a unique opportunity to acquire accurate information on near surface soil moisture. However, the newly developed algorithms for this mission need to be thoroughly tested on a wide range of land surface conditions and spatial resolutions, since their parameterization has been mostly limited to small scale field studies that focused on data from tower radiometers and simulation experiments. The SMOS mission is based on the relationship between the measured brightness temperature and the dielectric constant of the soil, which is related to its moisture content. Since this relationship is affected by a range of factors, including surface roughness and vegetation cover, the SMOS mission is relying on the multi-incidence angle observations to estimate some of the ancillary parameters required by the retrieval algorithm. In order to assess the performance of the core algorithm used by SMOS, multi-angle airborne data at L-band were studied from an Australian field campaign in 2005 (NAFE - National Airborne Field Experiment). The flights were conducted across three focus areas capturing a range of vegetation and soil moisture conditions on several observation days. The airborne instrument operated was the Polarimetric L-band Multi-beam Radiometer (PLMR) which provided dual-polarized brightness temperature measurements at six different incidence angles. The multi-angle observations were obtained by deploying a push-broom sensor rotated such that all beams were looking along the flight track, three forward and three backward, respectively. Corresponding ground sampling activities at specific focus farms included near surface-soil moisture, profile soil temperature, vegetation temperature, vegetation water content and biomass measurements. Additional rainfall, soil moisture profile and soil temperature data were collected at permanent monitoring sites nearby. In this research the L-MEB model was used to investigate the soil moisture retrieval results obtained when only specific ranges of multi-angle brightness temperature observations over wheat canopy were used. Moreover, the impact of varying moisture conditions and multi-parameter retrievals were studied to further assess the model performance and soil moisture accuracy. The results demonstrated overall good soil moisture estimates in relation to the ground measurements (ΔSM=0-0.03 m 3 /m 3 ), if i) soil moisture was retrieved solely and ii) all available brightness temperature data collected across 0-50° incidence angles were used. However when attempting the simultaneous retrieval of soil moisture together with ancillary data and focusing on specific angular ranges of observations, large variations in soil moisture estimates were observed. In particular, L-band data collected at incidence angles of 10° or higher seemed to be more strongly affected by the canopy, since vegetation effects of the dominant vertical wheat structure increased with larger incidence angles. Hence, L-band data measured at near-nadir views (0-10°) produced the soil moisture results closest to those measured - independently of the observed moisture conditions - when the optical depth and/or the surface roughness parameter were retrieved simultaneously with soil moisture.