Dissertations and Theses

Summary

– The accurate mapping of soil attributes presents a high cost, often unfeasible for agricultural use, mainly due to the large number of samples normally required for the collection and laboratory tests. In this sense, it is necessary to deepen research involving methodologies and new knowledge that have proven efficient in the purpose of generating soil maps with quality and possibility of cost reduction. For this purpose, the soil clay mapping was evaluated using magnetic susceptibility (MS) in the ordinary cokriging (OCK) interpolation procedure. The objective was to use the OCK method with the help of MS (analytical measurement obtained at a relatively lower cost than that of clay) to reduce the sampling density of the clay and generate more accurate maps. For this, the validation procedure was used with independent data (10% of the data) to verify the prediction errors (RMSE) that allowed to evaluate the accuracy of the maps. The studied area is a sugar cane field that has 870 ha located in Guatapará-SP. In this area measurements of soil MS and clay content were obtained from 371 samples collected at 0-0.25 m depth of a mesh with points separated by minimum distances varying from 145 m to 174 m. To evaluate the reduction of clay sampling density, 14 scenarios of different sample proportions of the two variables were tested. Sampling of the clay corresponded to densities of 1 point every 2,6; 3; 4; 5; 6; 7; 8; 9; 10; 11; 12; 13; 14 and 15 ha, that in each one was incorporated the sampled MS at 1 point every 2.6 ha. The results showed that this incorporation provided an increase in the accuracy of the estimates in 6 of the 14 scenarios studied. The scenario that presented the best performance, provided a 7.46% increase in accuracy with a 63% reduction in the sampling density for clay. The results of this study evidenced the potential of MS as an auxiliary covariate along with the OCK method in the mapping of clay with better accuracy and considerable sample savings.