Dissertations and Theses

Summary

Understanding soil erodibility is of paramount importance to delve into studies aimed at the reduction of soil erosion. The objective of this study was to investigate the efficiency of the use of magnetic susceptibility (MS) and diffuse reflectance spectroscopy (ERD) in pedotransfer functions that are capable of estimating the factors erodibility K and Ki latosols in the northeastern state of São Paulo. This study was conducted in 380 ha area under cultivation of cane sugar in Guatapará, São Paulo, Brazil. 86 samples in the "spike" strand (transect) and 150 samples were collected on the side. There were analyzes of size fractions, water dispersible clay (ADA), soil organic matter (OM), concentration of iron oxides, goethite ratio (Gt) and hematite (Hm) and between kaolinite (Ct) and gibbsite (Gb), magnetic susceptibility and calculated the parameters of erodibility K and Ki. The pedotransfer functions (PTF) to estimate the K and Ki factors as a function of SM, were modeled by linear regression analysis and model validation was performed by cross-validation. The SM was significantly correlated with K and Ki factors, probably because she was associated with the soil properties that govern the resistance of the soil to be eroded. Observing the Kriging maps, one realizes that changing the values of SM occurs in decreasing order towards the highest point to the lowest point, and the values of Kest and Kiest grow in the same direction. We conclude that the SM can be used as function Pedotransfer component in quantifying the erodibility oxisols the northeast region of the state of São Paulo, Brazil. To test the efficiency of the ERD in the prediction of K and Ki, 30 samples selected from 15 and 15 lateral transect were used. Spectroscopic used three bands (Vis-NIR-MIR) separately and combined. The spectral calibration was performed using chemometrics. The models were generated using the Partial Least Squares regression (PLS), using the cross-validation called leave-one -out. The precision and accuracy of modeling calculations were evaluated with the adjusted coefficient of determination (R²adj) and Residual Prediction Deviation (RPD), respectively. The importance of the variables in the prediction of the factors K and Ki was given by the index of importance of the variables (VIP). All properties studied, the spectroscopic system that estimated the same, was the MIR. Among the erodibility parameters, presented the best adjustments, was the Ki coefficient R²adj=0.57 and RDP=1.57. It was concluded that the use of the technique of ERD, especially in the MIR range, is effective in predicting the factors K and Ki latosols the northeast region of the state of São Paulo.