Radar Tomography for Remote Sensing of Forest Biomass and Water Dynamics

Doktorsavhandling, 2026

There is an increasing need for global monitoring of forests’ state and connections to climate. Synthetic aperture radar (SAR) observations can sense into dense canopies with high spatial resolution; they provide information on forest biomass and water status, important for quantifying the role of forests in the global carbon and hydrological cycles. Tomographic SAR (TomoSAR) techniques achieve vertical resolution, improving and expanding radar remote sensing capabilities.

Paper I investigates the sensitivity of P- and L-band (0.7 m and 0.2 m wavelengths) TomoSAR to above-ground biomass (AGB) in the TomoSense campaign. For the first time, differences between the AGB dependence of spruce and beech reflectivity profiles are identified. Moreover, unique observations of ground slope influence on AGB retrieval are presented, with significant effects for P-band. This is directly relevant for the exploitation of the recently launched BIOMASS satellite. Without separating forest types, AGB was estimated for P-/L-band (ground slopes below 10◦), with R² = 0.77/0.54 and RMSE = 11.4/12.0%.

New connections between radar observations and forest water dynamics are established. Paper II presents surprising signatures in L-band reflectivity of a forest subject to drought. A novel model is formulated, describing the reflectivity profiles, and model-estimated quantities strengthens connections to forest water dynamics; they show R² = 0.70 with vapor-pressure deficit, a driver of transpiration, and R² = 0.92 with soil moisture, indicating tree water stress. In Paper III, for the first time, tree hydraulics and tomographic radar observations are compared; uniquely, it is shown that reflectivity variations at P-band relate directly to crown hydraulic variations and at L-band are driven more by canopy attenuation. Paper IV presents a novel hydraulic forest model, and, for the first time, estimates forest transpiration using tomographic radar measurements.

Spaceborne TomoSAR imaging at L-band and above is limited by movements of canopy structures causing decorrelation between satellite passes. Paper V investigates novel radar correlation tomography techniques, which are robust against decorrelation. Importantly, these techniques may be implemented with tandem-formation SAR missions. This thesis strengthens prospects of TomoSAR for remote sensing of forest biomass and water dynamics, and contributes with techniques that facilitate new global observations.