*3.2. Analysis of the SSM Spatial Patterns*

To consistently analyze the spatial features of the SMAP and SMOS SSM maps at 1 km, their maps of daily differences were computed (SMAP\_AP1 minus SMOSL4) along the entire study period and the histogram of these daily SSM difference maps has been obtained, together with its mean and standard deviation (std). In addition, daily SSM difference maps have been temporally averaged and compared to the spatial distribution of the most common land cover types over the Iberian Peninsula.

Besides, taking into account different ancillary data (e.g., soil roughness, vegetation indices, skin temperature, or albedo) [24], high-resolution SMAP and SMOS SSM maps are derived from their respective TB, as described in Sections 2.1.1 and 2.1.2. However, there are noteworthy differences related to TB polarizations and incidence angles. While the SMAP disaggregation methodology in (1) uses one specific polarization (vertical) with a single incidence angle (40◦), the SMOS downscaling algorithm in (2) employs two polarizations (horizontal and vertical), and the average of three incidence angles (32.5◦ ± 5◦, 42.5◦ ± 5◦, and 52.5◦ ± 5◦) over the same target. To analyze the influence of TB data on the high-resolution SSM maps, the vertical SMAP L1C TB has been compared to the vertical SMOS L1C TB at the Earth's surface, using exclusively the central SMOS angle (42.5◦). To do so, the SMOS TB has been corrected by the geometry of the antenna, the ionospheric and atmospheric effects, linearly interpolated to the angles range 42.5 ± 5◦ and binned to a 25 km EASEv2 grid. The SMAP TB has been interpolated from the initial 36 km EASEv2 to the same grid of SMOS, using the nearest neighbor. Then, daily differences (SMAP minus SMOS TB) have been computed from April 2015 to December 2017. The coastal areas of the Iberian Peninsula were discarded to screen out the effect of sea-land contamination.

A low- vs. high-resolution study has also been performed to assess the variations, in volumetric units, between the original and the downscaled SSM maps of the same sensor. In this way, we assessed the impact of the different downscaling methods on spatial soil moisture patterns. To do this, coarse-resolution SMAP and SMOS maps were firstly interpolated to a 1 km grid using the nearest neighbor. The comparison was done by separately calculating the daily differences between the SMAP\_AP1 and the SMAPL2 maps, as well as the daily differences between SMOSL4 and SMOSL3 maps along the entire study period.
