Bayesian Data Fusion Framework for Soil Moisture Interpolation in Entre Ríos, Argentina: Analysis of Topographic Indices

The estimation of soil moisture contents on crop growth is the most important variable and, in many cases, determines yields. Many simulation models predict this variable by considering atmospheric conditions, crop water needs, and soil characteristics. In general, these models simulate soil moisture content for an average plant in the cultivated field without taking into account spatial variability. Relief is one of the characteristics that can explain this variability, so obtaining maps of soil moisture in the root zone has been addressed in this work through spatial interpolations obtained in sampling campaigns, together with the application of the Bayesian data fusion technique. In this work, soil moisture measurements were carried out in the first half of 2021 and the second half of 2022. With these data, several topographic indices were analyzed, finding that the inverse of the topographic wetness index and the digital terrain model best explain the spatial variability of soil moisture. Subsequently, data fusion techniques were applied by combining the results of the Ordinary Kriging interpolation method and these topographic indices. An analysis of the estimation errors was carried out using an independent set of data that did not participate in the spatial interpolations. It is observed that the application of the Bayesian data fusion method, considering these topographic indices, improves the soil moisture estimates compared to the use of the Ordinary Kriging interpolation method alone.