Towards a Remote Sensing-Based, Physically Consistent Soil Moisture Continuum: An Integrated Multi-Scale Framework from Surface to Subsurface 6 views

Soil moisture (SM) is a small component of the global water budget yet a major regulator of land–atmosphere exchanges, hydro-meteorological extremes, ecosystem functioning, and agriculture. However, four fundamental limitations still constrain current SM observing systems: (1) reliable information is largely confined to the upper ~5 cm, leaving the subsurface (20–50 cm) root zone poorly characterized at large scales; (2) vegetation water content and canopy structure strongly attenuate microwave signals, degrading retrievals over moderate to densely vegetated regions; (3) validation of satellite- and land surface model (LSM)–based products is hampered by uncertainties and spatial representativeness errors in in-situ reference data; and (4) a persistent spatial–temporal trade-off means that missions with fine spatial resolution often revisit infrequently, whereas frequently sampled missions operate at coarse footprints. To address these gaps in a coherent manner, this dissertation will advance four tightly coupled methodological components. First, a physics-guided subsurface SM estimation framework will be developed that uses diffusion-informed, depth-specific transfer times to extend satellite surface SM into the 20–50 cm root zone in a globally scalable and physically interpretable way. Second, a generalized two-step vegetation-mitigation scheme—combining system calibration with canopy attenuation correction—will be used to systematically reduce vegetation-induced distortions in L-band observations and improve SM retrievals in partially to densely vegetated landscapes. Third, validation and benchmarking strategies will explicitly incorporate spatial representativeness metrics to reconcile point-scale in-situ measurements with satellite footprints, yielding more robust assessments of satellite and LSM performance. Fourth, two complementary gap-filling frameworks, POBI and BSTI, will be rigorously compared and integrated to fuse their respective strengths in bias correction and spatio-temporal interpolation, thereby producing SM products that more effectively exploit existing observations while alleviating the spatial–temporal sampling deficiencies of current satellite missions. Collectively, these advances are expected to deliver a new generation of SM datasets and methods that provide deeper, more temporally and spatially consistent, and more trustworthy information for hydrologic prediction, climate diagnostics, and water-resource decision support. decision support.

PHD (Doctor of Philosophy)

English

2026-04-24