Evaluation of soil moisture products and Sentinel 1 & 2 data from THEIA platform for monitoring water status of agricultural plots and orchards in the Vaucluse department.

In the Mediterranean zone, the available water resources are subject to increasing tensions between users and call for a better assessment of the water consumption and use over territories, especially for agriculture. Sentinel missions now provide free remote sensing data with a high spatial and temporal resolution delivering regular information over large areas both on crop development and on the water status of various land surfaces. This study makes use of Sentinel data  for two main purposes : -i) to evaluate the accuracy of new soil moisture products obtained from Sentinel 1 & 2 delivered from the THEIA platform by using soil moisture data from the monitoring of an agricultural plot located in Avignon (France) where measurement are available from several years with different sensor types; -ii) to assess the potentialities of Sentinel 1 and 2 for monitoring soil moisture/irrigation in irrigated cherry orchards in the Ouveze basin (France). Results show that THEIA derived soil moisture values are significantly correlated with in situ measurements (at 5cm depth in soil), but with variation of the relationship between years, not linked to variation in soil roughness, leading to dispersion when all years are pooled (r² ~0.25-0.36) and under-estimation of higher water contents (>0.3 m3/m3). A normalization of signal data with the yearly amplitude could improve this correlation (r²=0.36-0.55). In relation with aim (ii), temporal profiles of spectral indices obtained for several orchards with Sentinel 2 allowed to clearly identify the trees’ phenology and the impact of the inter-row management. First results also showed a medium but significant correlation (r²=0.36) between the VV polarization extracted from Sentinel 1 data at the highest incidence angle and soil moisture measured under irrigated cherries orchards. These soil measurements, combined with spectral data on a longer time interval, available for two main irrigation practices (drip and microsprinkler) will allow for a better understanding of the input of irrigation water and its use by trees at the plot scale, useful for future modeling approaches of the water balance in orchards in relation with irrigation.