Integrating invasive and non-invasive monitoring sensors to detect field-scale soil hydrological behavior.

In recent decades, while great emphasis has been given to the monitoring of point-scale soil moisture patterns and field-scale integrated soil moisture, the measurement of matric potential has attracted little attention. Information on the soil matric potential is available in point-scale measurements but is still missing at field-scale. This state variable is necessary to understand hydrological fluxes and to determine the soil water retention function (WRF) for field-scale applications. In this study, we combine data from cosmic-ray neutron probes (CRNP, non-invasive proximal soil moisture sensors) and SoilNet wireless sensor networks (invasive ground-based soil moisture and matric potential sensors) installed in two sub-catchments with contrasting land-use (agroforestry vs. near-natural forest) to derive a field-scale WRF. We investigate the hypothesis that both sensor types provide effective measurements that are representative for the entire sub-catchment, as well as the drawbacks of integrating the different measurement scales of the sensor types (i.e., spatial-mean of distributed point-scale data vs. an integrated field-scale measurement). We found discrepancies in the data of the two sensor types related to the effects of the time-varying vertical measurement footprint of the CRNP, which induces a scale mismatch between CRNP-based soil moisture (referring mostly to near-surface depths) and the spatially averaged soil matric potential data measured at soil depths of 0.15 and 0.30 m. To remove the offsets, we opted to use the soil moisture index (SMI) based on the estimation of field capacity and wilting point, retrieved from the knowledge of the field-scale WRF. We found that the bimodality of SMI calculated with SoilNet-based soil moisture induced by Mediterranean rainfall seasonal behavior is not well-captured by CRNP-based soil moisture, except in a particularly dry year like 2017. The contrasts in SMI values between the two test sites were associated with differences in the spatial variability of soil moisture patterns explained by soil texture or terrain characteristics. We argue that field-scale WRFs are useful for the analysis of hydrological processes at the sub-catchment (field) scale and the application of distributed models