Identificación y evaluación de discrepancias entre los mapas de ambiente a partir de mapas de rendimiento

Precision agriculture uses various tools to optimize agricultural production, including environment maps for the differential application of inputs. However, maps generated from vegetative indices do not always adequately reflect areas of higher or lower yield, which can reduce the efficiency of input use. The objective of this final degree project is to identify and evaluate the discrepancies between environment maps generated using vegetative indices (NDVI, GNDVI) and corn and soybean yield maps. We worked with nine lots distributed in three areas of the provinces of Santa Fe and Buenos Aires, using yield maps from three consecutive harvests (27 maps in total) and environment maps provided by the MSU Agro company. The methodology included the development of a model in QGIS to filter outliers in the performance maps and compare them with the settings. Annual and average evaluations were carried out to analyze discrepancies between the maps. In addition to this, the kappa coefficient was calculated for each situation, and a principal component analysis (PCA) was performed in R to investigate possible relationships between the discrepancies and variables such as organic matter, nitrogen, and rainfall during the critical period. The results indicated that, on average, 52% of the points coincided with the expected performance, while 48% presented discrepancies, being more significant in low productivity environments. These areas showed greater interannual variability, possibly due to adverse climatic conditions. Furthermore, the kappa coefficient suggested that the use of several performance maps when setting, improves the precision of the settings, since they reduce the year effect. The PCAs revealed that the discrepancies could be influenced by soil variability and climatic factors. In conclusion, creating environment maps based on years with average climatic conditions is an effective strategy to reduce long-term discrepancies, although it is also important to consider soil variability to improve results.