Establecimiento de modelos parametrizados para estimación de posible presencia de contaminación y enfermedad del suelo mediante drones en la zona de Cayambe

Abstract

At present, agricultural production is essential to ensure food, as well as being a means of subsistence for several peoples within Ecuador and the world. However, the excessive application of fertilizers causes serious damage to the soil such as erosion, leading to a decrease in agricultural production. Given this problem, the present investigation proposes the estimation of soil quality parameters through the development of parameterized models that relate the elements such as nitrogen, phosphorus and potassium, through the use of unmanned aerial vehicles (UAV), and multispectral cameras, to find the places where excess or shortage of these nutrients is found. The models selected to make the prediction of soil nutrients are: to predict nitrogen was the EVI, with which an R2 of 0.712 and an RMSE of 0.38 was obtained, in terms of phosphorus the best index to make its prediction it was the NDVI, obtaining an R2 value of 0.858 and RMSE of 0.38, finally the model that best fits to make the potassium prediction is the NDVI with an R2 of 0.826 and an RMSE of 0.271. It is concluded that the variables studied are broadly related to the spectral indices, so that an estimate of the macronutrient content in the soil can be made correctly and with a relatively small error range.