Potential of hyperspectral remote sensing for field scale soil mapping and precision agriculture applications

Mapping within-field variation in soil properties opens up the possibility of employing variable agronomic management and precision farming technologies with potential environmental and economic benefits. However, the excessive cost of systematic direct soil sampling severely constrains the practical feasibility of site specific management based on soil variability information. Remote sensing offers a cost effective and efficient means for gathering a great deal of information on soil properties. The aim of the present work was to assess the potential of satellite hyperspectral imagery for the mapping of soil properties and the tilled layer of agricultural fields, in the context of precision agriculture applications. CHRIS-PROBA satellite images were acquired over two bare soil fields and their capability to provide estimates of soil texture and soil organic matter (SOM) at the field scale was assessed. Partial least squares regression (PLSR) models were developed on datasets spatially independent from those used for validation. Clay and sand could be estimated with intermediate accuracy, with values of RPD (ratio of performance to deviation) higher than 1.4. Root mean squared error (RMSE) values of 3.7 and 5.2 were obtained for clay in the two fields respectively. SOM estimates were not satisfactory, probably because of the limited range of spatial variation in the studied fields. Maps of uniform soil zones were obtained from measured and estimates soil texture data by means of fuzzy c-means classification. The resulting maps were then used for the parameterization of a simple water balance model, i.e. CropWat8.0, in order to simulate and compare uniform and variable-rate irrigation strategies. Simulation results suggest that site-specific irrigation allows to reduce significantly water losses by deep percolation, which occur when irrigation scheduling and volumes are calculated on the basis of average field soil properties. The present paper demonstrates the usefulness of satellite hyperspectral data for mapping soil spatial variability at the field scale, providing useful information for precision agriculture applications