Detecting changes in root zone soil moisture from radar vegetation backscatter
<p>Root zone soil moisture (&#952;<sub>rz</sub>) is a crucial component of the hydrological cycle and provides information for drought monitoring, irrigation scheduling, and carbon cycle modeling. During vegetation conditions, estimation of &#952;<sub>rz</sub> thru radar has so far only focused on retrieving surface soil moisture using the soil component of the total backscatter (&#963;<sub>soil</sub>), which is then assimilated into physical hydrological models. The utility of the vegetation component of the total backscatter (&#963;<sub>veg</sub>) has not been widely explored and is commonly corrected for in most soil moisture retrieval methods. However, &#963;<sub>veg </sub>provides information about vegetation water content. Furthermore, it has been known in agronomy that pre-dawn leaf water potential is in equilibrium with that of the soil. Therefore soil water status can be inferred by examining&#160; the vegetation water status. In this study, our main goal is to determine whether changes in root zone soil moisture (&#916;&#952;<sub>rz</sub>) shows corresponding changes in vegetation backscatter (&#916;&#963;<sub>veg</sub>) at pre-dawn. We utilized Sentinel-1 (S1) descending pass and in situ soil moisture measurements from 2016-2018 at two soil moisture networks (Raam and Twente) in the Netherlands. We focused on corn and grass which are the most dominant crops at the sites and considered the depth-averaged &#952;<sub>rz</sub> up to 40 cm to capture the rooting depths for both crops. Dubois&#8217; model formulation for VV-polarization was applied to estimate the surface roughness parameter (H<sub>rms</sub>) and &#963;<sub>soil </sub>during vegetated periods. Afterwards, the Water Cloud Model was used to derive &#963;<sub>veg</sub> by subtracting &#963;<sub>soil</sub> from S1 backscatter (&#963;<sub>tot</sub>). To ensure that S1 only measures vegetation water content, rainy days were excluded to remove the influence of intercepted rainfall on the backscatter. The slope of regression lines (&#946;) fitted over plots of &#916;&#963;<sub>veg</sub> against &#916;&#952;<sub>rz</sub> were used investigate the dynamics over a growing season. Our main result indicates that &#916;&#963;<sub>veg </sub>- &#916;&#952;<sub>rz</sub> relation is influenced by crop growth stage and changes in water content in the root zone. For corn, changes in &#946;&#8217;s over a growing season follow the trend in a crop coefficient (K<sub>c</sub>) curve, which is a measure of crop water requirements. Grasses, which are perennial crops, show trends corresponding to the mature crop stage. The correlation between soil moisture (&#916;&#952;) at specific soil depths (5, 10, 20, and 40 cm) and &#916;&#963;<sub>veg </sub> matches root growth for corn and known rooting depths for both corn and grass. Dry spells (e.g. July 2018) and a large increase in root zone water content in between two dry-day S1 overpass (e.g. from rainfall) result in a lower &#946;, which indicates that &#916;&#963;<sub>veg</sub> does not match well with &#916;&#952;<sub>rz</sub>. The influence of vegetation on S1 backscatter is more pronounced for corn, which translated to a clearer &#916;&#963;<sub>veg</sub> - &#916;&#952;<sub>rz</sub> relation compared to grass. The sensitivity of &#916;&#963;<sub>veg</sub> to &#916;&#952;<sub>rz</sub> in corn means that the analysis may be applicable to other broad leaf crops or forested areas, with potential applications for monitoring&#160; periods of water stress.</p>