Precision agriculture: A spatial econometric analysis of a Portuguese maize yield

Nowadays, maize is the most important cereal in the world and its production has been increasing both worldwide and in Portugal, over the years. The constant technological development has led to the creation of new techniques such as precision agriculture, to better meet the global needs of this primordial cereal as well as optimize its production. This research was developed jointly with the firm Agro Analítica from the agriculture sector, whose area of expertise is Precision Agriculture and System Optimization. In this manner, the present work aims to estimate a function for the maize yield identifying the relevant determining factors, and their effect, on maize productivity on an exploitation of a firm in Azinhaga, Golegã, district of Santarém, Portugal for the year 2020. Using appropriate software, this dissertation applies the most recent spatial econometric methods to cross-sectional data, in order to properly include spatial dependence in the estimation. Thus, the appropriate models were estimated: Spatial Lag Model (SLM), Spatial Error Model (SEM) and SARAR(1,1) Model, whose use was recommended by the diagnosis to OLS (Ordinary Least Square) residuals. The elected model was the SARAR(1,1), capturing the spatial dependence and heteroscedasticity in the data, with an accuracy of approximately 90%. In this framework, it was concluded that maize yield, in the year and area under study, is positively influenced by factors such as the sowing density, applied sulfur trioxide (SO3) and a specific variety of seed. Regarding the fertilization, nitrogen and potassium, and irrigation of the crop, presented a non-linear (quadratic) relationship with the maize yield. Also influencing the yield, there are weather-related variables measured by stage of the maize life cycle, that proved to be significant at explaining the variable under study such as the relative humidity, the temperature, and the wind velocity.