Monitoring soil–plant interactions and maize yield by satellite vegetation indexes, soil electrical conductivity and management zones.

In modern agriculture, understanding the spatio-temporal variability in crop fields and the implications of environmental factors in soil management are important for sustainable practices. In this case, the management zones (MZ) can aid agricultural practices by indicating locations in a crop field where the production might be restricted and requires a specific management. Currently, there are many datasets and methodological strategies for designing MZs, and the outcomes from these methods are quite different. Therefore, this research aims to compare the performance of different remote/proximal sensing inputs to retrieve data from soil and plants, define their relationship with corn yield, and the potential of these datasets to design MZs. The study was conducted at a corn field located in Paraguay. The datasets used in our methodology corresponded to (I) electrical conductivity, (II) soil data from conventional laboratory analysis and (III) spectral information (optical and thermal) derived from Landsat 8 images. MZs were generated from each one of the datasets and later they were compared to yield maps. In this case, zoning performances were evaluated by the similarity between MZs and yield maps. The best results were achieved with the spectral vegetation indices from Landsat 8. Correlations between vegetation indexes and yield reached a maximum value of 0.75 for NBR2 index, but EVI, SAVI and NDVI also presented good results (r > 0.7). Furthermore, vegetation indexes of corn at V8 phenological stage provided the best agreement between MZs and yield. Finally, the MZs derived from spectral data could define yield-limiting zones inside the crop field. [ABSTRACT FROM AUTHOR]