A four-dimensional soil moisture product: closing the soil profile gap with a SWI-SMAR approach.

The increasing frequency and intensity of extreme meteorological and climatic events, such as droughts and floods, represent a direct manifestation of the growing impacts of climate change on natural hazards. Within the context of these phenomena, one of the crucial parameters to be considered is the variability of the Soil Moisture (SM), namely the soil water content from the soil surface up to the root zone (RZ) depth. However, to fully understand the four dimensions (i.e., 3D spatial plus temporal) SM evolution and its implications for associated risks, it is essential to have access to high-resolution temporal and spatial data. Currently, SM information retrieval relies on three main approaches: ground-based measurements, hydrological modeling, and remote sensing. However, each method has its own limitations as well as advantages. Ground-based measurements often provide high quality point data, but they are unevenly distributed across territories with limited spatial extension, and that can be affected by systematic probe errors. On the other hand, hydrological modelling can extend the scale of application but relies heavily on the fundamental assumptions of the model used as well as on the quality of the available input/ancillary data. Finally, satellite sensors can provide a synoptic view and high frequency of observation, but often struggle to simultaneously provide adequate spatial and temporal resolution. The integration of data from ground measurements, remote sensing, and hydrological modeling could allow for filling the gap(s) in each approach, fostering the creation of an innovative and cost-effective monitoring system. Such an integrated approach would also be capable of providing precise measurements along the soil profile, addressing another open field in SM research. This thesis aims at bridging these gaps by addressing three key questions. Firstly, it seeks to explore the peculiarities of SM products within the European context encompassing the Mediterranean area recognized as one of the areas most impacted by climate change. Secondly, it aims to investigate the possibility of integrating different SM products on a large scale. Lastly, it focuses on examining the feasibility of constructing high-resolution spatial (x, y) and temporal (t) information at various soil depths (d). The development and testing of a 4D SM product would facilitate the acquisition of accurate SM information throughout the soil layer, from the surface to the root zone, holding the potential to significantly enhance our understanding of soil moisture dynamics. To reach such results, several intermediate steps have been developed, starting with a preliminary analysis of the accuracy of a few satellite-based SM products, trying to investigate which one offers the best performance also concerning the goal of create a 4D SM product. Such an analysis has been carried out, for the first time in this work, at the European ecoregions spatial scale, by an intercomparison of five SM datasets with the ground information made available by the International Soil Moisture Network (ISMN). This preliminary analysis allowed us to both assess the different performance of the considered products, and to understand that the ecoregion scale could be a suitable investigation level to capture dynamic behavior patterns. A deeper exploration of previous findings and a confirmation of the behavior of these ecoregions concerning seasonality, including the complete removal of seasonality and studying the relationships with phenological phases, were then carried out. Finally, focusing only on active microwave sensors, which have demonstrated a different level of accuracy, the 4D SM product has been developed and tested. In particular, ASCAT (H119-H120) and the enhanced S-1 SM product, both based on active measurements in the C-band, were considered. The first is sub-daily provided at about 25km of spatial resolution, while the second one can allow for information at a higher spatial resolution (1km for the Surface Soil Moisture – SSM - data) but with a sub-weekly temporal resolution. Two different approaches, the Soil Water Index (SWI) and Soil Moisture Analytical Relationship (SMAR), respectively, were then applied to the data, first to produce the blended SSM product, and then the RZSM one. The results obtained provide an overview of the fundamental role of SM, which is helpful in issues related to climate change. In particular, the application of SCAT-SAR SWI SMAR demonstrated a substantial correlation with in-situ data (r ~ 0.8),along with significantly reduced prediction errors (RMSD ~0.003/0.01). At the regional level, the application of SMAR provided more consistent information on RZSM with real hydraulic processes compared to the SWI application, which displayed a simple reduction in saturation values while maintaining the same input data pattern. This discrepancy can be attributed to the absence of a clear connection between soil depth and the parameter T. The proposed method involves a limited number of parameters and is easily implementable. Since the SWI-SMAR approach is based on a recursive algorithm, the output improves as the time series length increases. This approach could prove particularly useful for large-scale studies to advance our understanding of the effects of climate change and risk management. However, current limitations arise from approximations of parameters as the normalized coefficient of losses (a), or the normalized coefficient related to the soil properties and the depth (b), or the saturation at wilting point (sw), or the saturation at field capacity (sc), and the computational capacity required for processing large volumes of data (with a nine GB output for the sole application in the Basilicata region). Addressing these obvious gaps and enhancing the accuracy of the predictions along the soil profile would require improvements in parameter derivation. Future developments could include integration with additional field experiments and the use of artificial intelligence methodologies.