A NOVEL DOWNSCALING METHODOLOGY FOR INTERMEDIATE RESOLUTION RADIOMETER DATA FOR SMAP
IGARSS 2015 – Remote Sensing: Understanding the Earth for Safer Word – July 26-31, Milan, Italy
Abstract:
Forthcoming NASA’s Soil Moisture Active Passive (SMAP) mission schedule for launching on early 2015 will provide simultaneous L-band radiometric brightness temperature (T b) and synthetic aperture radar (SAR) backscattering (σ) on a global basis every 3 days. Exploiting the combination of both active and passive microwave observations poses a challenge for developing robust soil moisture (mv ) retrievals. L-band radiometer observations are characterized by their sensitivity to mv but have as a disadvantage their low spatial resolution (36 km in the case of SMAP). On the other hand, backscattering coefficient measured by SAR systems have higher spatial resolution (∼ 3 km in SMAP) but show greater sensitivity to soil roughness and vegetation, and is affected by speckle noise, which makes mv retrieval more difficult. Synergy between active and passive observations seeks to benefit from these features. Baseline SMAP active/passive mv retrieval consists on T b disaggregation at a medium resolution scale (9 km) based on a linear coefficient (β) that relates σV V to variations of T bv and a coefficient Γ that projects σHV variations to σV V domain. Both coefficients are estimated statistically at coarse resolution using multi-temporal active and passive measurements. Once T b is downscaled using this disaggregation algorithm, it is then used for mv retrieval using Single Channel Algorithm [1]. Other combined active/passive (CAP) approaches proposed include change detection approaches [2, 3], a CAP with joint-physics [4] and SMAP optional algorithms: T b disaggregation at high (3 km), medium (9 km) resolution [1] or mv disaggregation [5]. The algorithm detailed in this work attempts to disaggregate T b at medium resolution by addressing two key issues:
coupling an active and passive semiempirical models to derive an analytical relation between T bv and σV V and (2) merging radar-derived and radiometer-based T bv at different spatial resolutions.