| Abstract: |
The NASA Soil Moisture Active-Passive (SMAP) mission aims to disaggregate low-resolution L-Band radiometer observations (36 km) with higher resolution radar observations (1-3 km) to obtain an intermediate resolution soil moisture product (1-9 km) [1]. The downscaling algorithm of SMAP is based on the assumption that covariations of radar and radiometer observations follow a linear functional relationship, which is captured by a slope parameter β. It is understood that β reflects local vegetation cover conditions and the downscaling accuracy of SMAP is strongly dependent on the estimation of β [2].
Since SMAP’s L-Band radar stopped operations in July 2015, a substitution with Sentinel 1’s C-Band radar for the operational disaggregation of SMAP L-Band brightness temperatures is in preparation [3]. To assist the novel SMAP-Sentinel 1 product, further research on L-Band radiometer and C-Band radar signal covariations is needed, especially concerning the stronger vegetation influence on C-Band signals with respect to L-Band.
This research study investigates vegetation effects on covariations of radiometer and radar observations, comparing the original SMAP (L/L-Band) and novel SMAP-Sentinel 1 (C/L-Band) frequency configurations. More specifically, simulations of the linear slope β between radiometer and radar observations are carried out for three distinct vegetation types (corn, wheat, coniferous forest) and a wide range of vegetation water content (VWC) conditions (wheat: 0-2.5 [kg/m²], corn: 0-6.5 [kg/m²], forest: 4.5-14.5 [kg/m²]). The respective active and passive microwave signatures are obtained from extensive simulations with the Tor Vergata scattering and emission model [4], utilizing well-validated parameter sets. To gain a physical understanding of vegetation effects on L/L-Band and C/L-Band signal covariations, radar signatures are decomposed into elementary scattering mechanisms (surface, double bounce, multiple scattering) and their influence on β is assessed. Finally, random signal disturbances are added to the simulated microwave signatures, and the resulting effects on β are captured by an errors-in-variables model.
For all vegetation types (corn, wheat and coniferous forest), we find that the slope β between radiometer and radar signals is biased towards zero with increasing vegetation cover due to a combination of vegetation and signal disturbance effects. This trend is in accordance with the results of previous studies for the L/L-Band case [4]. The main driver for the above-named effect is a decrease of radar sensitivity to soil moisture with increasing vegetation cover, which more severely affects the C-Band radar with respect to L-Band. Consequently, strong C/L-Band active-passive signal covariations are expected until maximum VWC values in the range of 1-6 [kg/m²] (depending on vegetation type and polarization), as opposed to 4-8 [kg/m²] for the L/L-Band case. In conclusion, although C/L-Band signal covariations are more adversely affected by vegetation cover, promising results are obtained for both the original SMAP and novel SMAP-Sentinel 1 configuration.
[1] Entekhabi, D., et al., (2010). The soil moisture active passive (SMAP) mission. Proceedings of the IEEE, 98(5), 704-716.
[2] Das, N. N., Entekhabi, D., Dunbar, R. S., Njoku, E. G., & Yueh, S. H. (2016). Uncertainty Estimates in the SMAP Combined Active–Passive Downscaled Brightness Temperature. IEEE Transactions on Geoscience and Remote Sensing, 54(2), 640-650.
[3] Das, N. N., Entekhabi, D., Kim, S., Jagdhuber, T., Dunbar, S., Yueh, S., & Colliander, A. (2017). High-Resolution Enhanced Product based on SMAP Active-Passive Approach using Sentinel 1A and 1B SAR Data. In Geoscience and Remote Sensing Symposium (IGARSS), 2017 (pp. 2543-2545). IEEE.
[4] Guerriero, L., Ferrazzoli, P., Vittucci, C., Rahmoune, R., Aurizzi, M., Mattioni, A. (2016) L-band passive and active signatures of vegetated soil: simulations with a unified model, IEEE Journal of Selected Topics in Earth Observations and Remote Sensing, vol. 9, pp. 2520-2531. |
