2021-12-15 REMOTE SENSING OF ENVIRONMENT 2021 267(卷), null(期), (null页)
Surface soil moisture and vegetation optical depth (VOD), as an indicator of vegetation wet biomass, from passive microwave remote sensing have been increasingly applied in global ecology and climate research. Both soil moisture and VOD are retrieved from satellite brightness temperature measurements assuming a zeroth order radiative transfer model, commonly known as the tau-omega model. In this model the emission of a vegetated surface is dependent on soil moisture, vegetation absorption and vegetation scattering. Vegetation scattering is normally represented by the single scattering albedo, omega, and is commonly assumed to be a time-invariant calibration parameter to achieve high accuracy in soil moisture estimation. Therefore, little is known about omega dynamics in the context of its ecological information content. Furthermore, VOD and omega are functions of more fundamental absorption and scattering coefficients kappa(a) and kappa(s). In this study, we retrieve both VOD and omega as well as kappa(a) and kappa(s) in two separate tau-omega model frameworks using known soil moisture information. Our sensitivity analysis confirms that vegetation attenuation has a strong impact on the sensitivity of the retrieval to noise. If vegetation attenuation approaches zero, omega is weakly constrained, which leads to strong w dynamics. If vegetation attenuation is very high, VOD, kappa(a) and kappa(s) are increasingly less constrained and susceptible to noise, while omega becomes quasi time-invariant. Coinciding with our sensitivity study, global retrievals from SMAP brightness temperatures exhibit large omega dynamics in drylands and deserts, decreasing rapidly with increasing vegetation cover. In drylands omega might peak during the dry season at the seasonal VOD minimum. With increasing vegetation attenuation, omega dynamics start to follow leaf phenology before they transition to a nearly time-invariant omega in forested areas. VOD, kappa(a) and kappa(s) follow the dynamics of wet biomass and peak after the maximum of leaf area index and precipitation. Our results suggest that zeroth order scattering is generally time-invariant over a majority of the global vegetated areas. In general, time dynamic omega and kappa(s) might provide additional information on the state of the vegetation canopy, which can be valuable for biosphere studies and relevant for the parameterization of vegetation scattering in soil moisture retrievals.