Multiple Scattering Effects with Cyclical Correction in Active Remote Sensing of Vegetated Surface Using Vector Radiative Transfer Theory.
Tien-Hao Liao,Seung-Bum Kim,Shurun Tan,Leung Tsang,Chenxin Su,Thomas J. Jackson
DOI: https://doi.org/10.1109/jstars.2015.2505638
IF: 4.715
2016-01-01
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Abstract:The energy transport in a vegetated (corn) surface layer is examined by solving the vector radiative transfer equation using a numerical iterative approach. This approach allows a higher order that includes the multiple scattering effects. Multiple scattering effects are important when the optical thickness and scattering albedo of the vegetation layer are large. When both the albedo and the optical thickness exceed 0.4, higher orders contribute significantly (e.g., vertical polarization at L-band). The approach is applied to vegetated surfaces using typical crop structure for backscattering from L-band to Ku-band. For corn fields at L-band, multiple scattering effects are more important for vertical scattered wave with vertical incidence (VV). For example, when vegetation water content (VWC) is 3kg/m2, the deviation between first order and multiple scattering for corn field for VV could be 3.5 dB while 0.7 dB for horizontal scattered wave with horizontal incidence (HH). The iterative approach also allows the separation of the contribution to backscattering from each scattering order and scattering mechanism. Each scattering mechanism is associated with a unique scattering path. By examining the duality of the paths, we are able to identify the cyclical terms with existence of a reflective boundary. The cyclical correction to the backscattering accounts for backscattering enhancement effects on the copolarization by a factor of two. The approach is validated against the SMAPVEX12 L-band corn dataset over the entire crop growth and large soil moisture variations. The model prediction matches the observation with 1.93 and 1.46 dB root-mean-square error (RMSE) for VV and HH, respectively, while correlations are 0.67 and 0.88, respectively. Time-series retrieval is also applied successfully for both soil moisture and VWC with 0.06 cm3/cm3 and 0.44 kg/m2 RMSE, respectively, while correlations are 0.7 and 0.92, respectively. For large VWC, this approach corrects the underestimated backscatters in the single scattering caused by large attenuation.