Dissertations and Theses
Uncertainties in the estimation of the spatial variability of the CO2 emission of the soil and edaphic properties in the area of raw cane.
Author: Daniel De Bortoli Teixeira
Keywords: soil respiration, geostatistics, ordinary kriging, Gaussian sequential simulation.
Summary
The soil CO2 emission (FCO2) presents high spatial variability, due to the great spatial dependence on the soil properties that influence it. The objective of this study was to characterize and relate the variability and spatial distribution of FCO2, soil temperature, water free porosity (PLA), soil organic matter (OM) and soil density (Ds), ( ii) to evaluate the accuracy of the results provided by the ordinary kriging (KO) and Gaussian sequential simulation (SSG), and (iii) to evaluate the uncertainty in predicting the spatial variability of FCO2 and other properties using SSG. The study was conducted in a regular sample mesh of 60 x 60 m2 with 141 points, with minimum spacing ranging from 0.50 to 10 m, installed in sugarcane area. In these points the FCO2, soil temperature, PLA, determined based on the average of 07 days of evaluation, MO and Ds, were evaluated. All variables presented a spatial dependence structure, being adjusted Gaussian, spherical and exponential models. The configuration of the sample mesh and possibly the presence of thick layer of crop residues on the soil influenced the spatial variability structure of FCO2, temperature and MO. FCO2 showed positive correlations with the MO (r = 0.25, p <0.05) and PLA (r = 0.27, p <0.01) and negative correlation with Ds (r = -0.41, p < 0.01). However, when the spatially estimated digital values (N = 8,833) are considered, the PLA becomes the main variable responsible for the spatial characteristics of the FCO2, presenting a correlation of 0.26 (p <0.01). The individual simulations provided better replication of cumulative distribution functions (fdac) and variograms compared to KO and E-type estimation for all analyzed variables. The results of the analyzes demonstrate strong similarities between the E-type estimates and those generated by the KO procedure, for all the estimated properties. The greatest uncertainties in FCO2 prediction were associated with the regions of the studied area with the highest observed and estimated values, producing estimates, during the studied period, of 0.18 to 1.85 t of CO2 ha-1 depending on the different simulated scenarios. For PLA and Ds regions with lower sample density produced greater uncertainties. The accounting of uncertainties associated with the estimation of soil properties helps to understand the possible FCO2 patterns. Knowledge of the uncertainties generated through the different estimation scenarios can be included in inventories of greenhouse gases resulting in more conservative estimates of the emission potential of such gases.