Prediction of Soil Carbon Stock in the PIAUI State Coast by Remote Sensing

The objective of this study was to determine the organic carbon content and soil carbon stocks under different vegetations in the Parnaíba River Delta (PRD), located in the Brazilian state of Piauí, and to estimate them by using three predictive methods and the spectral bands and vegetation indexes derived from the Landsat 8 images. The study was carried out in the Parnaíba River Delta Environmental Protection Area (APA), Piauí, on the coast of northeastern Brazil and the domain of the Caatinga, in areas with five different vegetations: psammophile pioneer vegetation, dune subevergreen vegetation, mangrove evergreen vegetation, floodplain vegetation and vegetation associated with Carnaubals. Soil samples were collected for 40 points distributed in the area, where carbon concentrations and carbon stocks were determined. The Tukey test at 5% probability evaluated vegetation’s interaction with soil variables. Afterward, the carbon concentration of 0–10 cm (SOC), and carbon stocks of 0–30 cm (CS30) and 0–100 cm (CS100) were predicted using three predictive methods: multiple linear regression (MLR), ordinary kriging (OK), and regression kriging (RK). The results r that the soils under the mangrove evergreen vegetation showed the highest averages for carbon concentrations and carbon stocks for all stratus. In the SOC, CS30 and CS100 predictions, it was observed that RK had the lowest RMSE (5.54 g.kg−1, 11.70 Mg ha−1, and 38.35 Mg ha−1, respectively) and highest R2 (0.97, 0.89, and 0.95, respectively), being considered the best method for predicting these variables in the study area.