*2.1. REMEDHUS Soil Moisture*

REMEDHUS is a soil moisture network located in the central part of the Duero basin in Spain and covering an area of approximately 1300 km<sup>2</sup> (41.1–41.5◦N, 5.1–5.7◦W). This region has a continental semiarid Mediterranean climate characterized by a clear water deficit, especially during summer. The location of the network was chosen precisely for these criteria, i.e., to monitor crop behavior under the water-limited conditions and over land covers representative of the region. In REMEDHUS, the main land use is agricultural, with rainfed crops such as cereals, legumes, and vineyards. These crops are adapted to the scarcity of water because most of them are fed only by precipitation, which is approximately 381 mm/year on average [45]. The soils are mainly sandy, leading to a limited water holding capacity.

Two sensor types are installed in REMEDHUS. First, there are 22 stations measuring SSM (0–5 cm) using Hydra Probes (Stevens Water Monitoring, Inc., Portland, OR, USA). Out of the 22 stations, 14 also measure soil moisture along the soil profile using EnviroSMART (Sentek Pty. Ltd., Stepney, SA, Australia) sensors deployed at 25, 50, and 100 cm depths. These depth levels ensure a complete measurement of the soil moisture at the root zone in the area [43].

For the assessment of the ATI-derived SSM, the in situ SSM data provided by all 22 stations in REMEDHUS were used. For the analysis of the RZSM estimates, the soil moisture observations at surface and root zone level from the 14 stations were used. Since the observations are on an hourly basis, soil moisture measurements at each station were first daily averaged. A representative estimate of in situ RZSM was obtained for each station by averaging the daily soil moisture measurements at the different depths (5, 25, 50, and 100 cm).

#### *2.2. SMAP L4 Soil Moisture*

Among the variety of SMAP products available, the global SMAP L4 soil moisture geophysical data over a 9-km Equal-Area Scalable Earth (EASE)-2 grid [46] were selected. The SMAP brightness temperature—originally at coarser resolution—was downscaled to 9 km using the Backus-Gilbert optimal interpolation. The SMAP L4 soil moisture was derived by assimilating this brightness temperature into the NASA catchment land-surface model, which interpolates and extrapolates the SMAP observations in time and in space [47]. The NASA model describes the vertical transfer of soil moisture between the surface and the root zone, built on a set of Richard's equation calculations under unsaturated conditions. This model is driven by observation-based surface meteorological forcing, including precipitation [30].

The SMAP L4 product provides SSM (0–5 cm) and RZSM (0–100 cm) estimations at a 3 h temporal resolution. In this study, the area corresponding to the Iberian Peninsula (34◦N–45◦N, 11◦W–5◦E) was clipped from the global SMAP L4 soil moisture maps. The resulting maps of SSM and RZSM were daily averaged.
