Linking Crop Models with Highly Resolved Soil Sensor Observations to Improve Spatial Simulation of Soil-Crop Interactions.

Soil properties often show a high intra-field variability, which requires an adjusted site-specific agricultural management on field-scale to avoid environmental pollution and to enhance resource use efficiency. Process-oriented crop models like the HERMES model, are able to provide variable outputs, but the quality of site-specific model results strongly depends on the quality of required input variables. As soil investigation with a high resolution in space and time is expensive and time-consuming, different sensor systems are calibrated and tested in the project network Intelligence for Soil, funded within the national programme BonaRes. Sensors may provide highly resolved soil information driving the models and state variables to actualize simulated states. First investigations concentrate on a well-documented agricultural field located in Westphalia, characterised by a high spatial variability of soil texture, ranging from sand to loam. Soil investigation has been conducted at 60 grid points providing spatial variable inputs for HERMES. Simulation quality was validated by available values of nitrogen and soil water content at each grid point in spring and after harvest of wheat. In a second step, a highly resolved data set of soil electrical conductivity (ECa) is merged with the texture information at each grid point and the relation is analysed inter alia by multiple linear regression. High coefficients of determination allow the calculation of ECa-based texture information, which is essential for running HERMES. Due to the high quantity of measured ECa values on the field, subsequent simulations on varying initial texture grids are executed. As a result, more differentiated spatial patterns of soil inventory, simulated nitrogen and water supply and simulated yields are generated. Compared with the pattern of observed yields, model improvement by implementation of sensor information is achieved and determination of small-scale, soil induced patterns of nitrogen efficiency and yield potential becomes possible.