Precision agriculture in sugar cane: instrumentation of a harvester, yield and soil attributes mapping

Author(s)

Domingos Guilherme Pellegrino Cerri

• Document type: Doctoral Thesis
• Press: Campinas, SP.
• Institution: Universidade Estadual de Campinas (UNICAMP). Faculdade de Engenharia Agricola
• Defense date: 2005-05-17

Examining board members

Paulo Sergio Graziano Magalhães; Nelson Luis Cappelli; José Paulo Molin; Jose Eduardo Cora; Jose Ruy Porto de Carvalho

• Advisor: Paulo Sergio Graziano Magalhães

Abstract

The present research aims to instrument a sugarcane harvester in order to obtain the crop yield maps and, at the same time, evaluate and correlate the physical and chemical attributes of the soil using the yield data. In order to obtain these maps, an area measuring 42 ha located at São João Sugar and Alcohol Mill in Araras, São Paulo, Brazil. The yield mapping system of sugarcane crop developed (SIMPROCANA) uses load cells as an instrument to weighing the harvested material and it is capable to measure the flow passing by the matting before being launched to the transport vehicle. These data, together with the information obtained by a Global Positioning System (GPS) installed on the harvester, allow the elaboration of a digital map with the help of a Geographical Information System - GIS that represents the production on the surface for the harvested area. This system was tested under laboratory and field conditions. SIMPROCANA represented satisfactorily the spatial variability of the crop yield. The weight obtained by this system compared with that of sugarcane load in a mill's scale presented an average error of 0.96%, with a maximum of 6.4%. The result is not satisfactory anymore when it compares the weight obtained by using SIMPROCANA with that from the transportation vehicle weighed in the field. The average error associated in this case is around ten fold higher, when the association with the weight obtained in the mill's scale is performed. In relation to the making of map of physical and chemical attributes of soil (granulometry, macro and micronutrients, acidity, density, and resistance to penetration), two samplings were made in consecutive years. In the first sampling, 170 samples were collected in a 42 ha area in a regular grid of 50 x 50 m, helped by a sampler vehicle. In the second one, 49 samples in a 100 x 100 m grid defined by the maximum reaching obtained by geostatistics analysis performed at the first sampling. The results of the first sampling were correlated with the sugarcane yield obtained by means of weighing the transport vehicle in the field. This alternative method allowed the using of the results from the first sampling once the harvester was not properly instrumented. The data from the second sampling were related to the yield mapping obtained by using SIMPROCANA. The results of the physical and chemical analyses used for the construction of semivariograms models adjusted better at the 50 x 50 m grid when compared to that of 100 x 100 m. the correlations between the yield and the soil attributes presented lower values. The higher indexes occurred for the variable altitudes (- 0.47%), clay rate (0.51%), and copper (0.55). Despite these low correlations, it was possible to recommend a varied rate application of inputs based exclusively on the information obtained from the chemical attributes of the soil. Simulation tests showed that the application costs under a varied rate of phosphated and potash fertilizers are practically half of the value calculated for a fixed rate. In relation to the liming, there was not significative variation in the prices of lime application spread by both methods (AU)