Espacialização da densidade do solo e do estoque de carbono da Bacia Guapi-Macacu/RJ, determinados por meio de funções de pedotransferência (FPTs)

Abstract

This study aimed to model the soil attribute bulk density (BD) by pedotransfer function (PTFs), spatializing it and afterwards mapping the carbon stock (CS) in Guapi-Macacu watershed, in state of Rio de Janeiro, based on Digital Soil Mapping (DSM) methodology. The data used is derived from a high intensity soil survey carried out by ‘’EMBRAPA Solos’’ during the last quarter of 2010 in the area. Thus, 99 soil profiles were sampled and analyzed, and of those, 66 had bulk density values determined. To validate the PTFs developed in the process, comparison with published functions was made from 13 samples that had been reserved. Two bulk density functions were derived from multiple linear regression and nonlinear regression, using 16 variables and relying on the statistical software Statgraphics Centurion XVI. Stepwise procedures and Comparison of Alternative models of the related program helped to choose the best models which have greater simplicity, normal residuals, larger R2, and minimum error. Nevertheless, the non-linear model, the one which involves the single organic carbon (OC) variable and which is easier to be determined, was used for spatialization of BD. This function allowed assessing BD value for the other soil profiles. Geostatistical analysis (semivariogram) performed from GS+ 9 program, showed spatial correlation of A horizon thickness, while OC exhibited high randomness. Since, A horizon thickness was spatialized through kriging interpolator, OC interpolated by the deterministic interpolator IDW (Inverse Distance weight) and BD mapped by using his pedotransfer function through Raster Calculator tools of ArcGIS v. 10. Bulk density varied from 0.87 and 1.95 Mg.m-3. Finally, the estimated carbon stock (CS= Ds.C.E/10, Mg.ha-1) range was 9.2 to 212.2 Mg.ha-1, and was spatialized following the same bulk density process. Variance analysis of CS mapped regarding land use and slope classes were performed by Fisher LSD test. The results indicated that among the different land use classes, only pasture and altitude vegetation are significantly different at 5% risk. For the six slope classes, has been observed significant difference exclusively between flat land and steep area at 5% risk