Dissertations and Theses

Summary

Gas emissions from agricultural activities can not be differentiated from fossil fuel combustion, since both are also sources of greenhouse gas production. In this way it is necessary to characterize the main aspects of the emission of this gas in the soils quantitatively, taking into account its spatial variability inclusively. In this work, we determined among the methods of ordinary kriging (KO) and Gaussian sequential simulation (SSG) which obtained a more accurate representation of the properties evaluated in the field. The experimental area was established in a red clayey eutrophic Latosol with no vegetation cover, located in the municipality of Jaboticabal, SP, FCAV-UNESP Campus (21º15 'South 48º18' West). Before the beginning of the evaluation of the CO2 emission of the soil, the area was meshed with the purpose of exposing the organic matter, which was protected inside the aggregates, the decomposition (oxidation) carried out by the microorganisms present in the soil. At the study site, on the uncovered vegetation soil, 64 points were installed, forming a regular grid with separation distance between points of 5 m (minimum) and 10 m (maximum) covering a total area of 110 x 15 m. The emission was quantified through the use of a ground chamber coupled to the portable CO2 analysis system (Model 8100) manufactured by LI-COR, Nebraska, USA. At the same time, the soil temperature was evaluated using a sensor coupled to the LI-8100 photosynthesis and humidity (% volume) with TDR-Campbel®, both in the 0-0, 20 m. Soil samples (0-0.20 m) were also collected at each grating point for determination of soil physical and chemical properties. Our results indicate that the afternoon period presents a higher emission (4.54 μmol m-2s-1) than in the morning (6.24 μmol m-2s-1), but both periods presented similar variabilities, with high CV values. For the evaluation of the best interpolation method for each variable, the external validation technique was used, where the actual values observed in 5 grid points were compared with the estimated value in these locations through the semivariogram adjustment. This comparison was made by adding the square of the residuals (SQR), comparing the observed and expected values. SSG was more efficient in estimating non-sampled values for soil CO2 emission in both periods, showing higher correlations between KO and SSG maps in the afternoon (R2 = 0.99). The temperature and humidity of the soil in the morning, sand, DMP, Ca and Mg were estimated more efficiently by the SSG technique. In contrast, kriging provided better results for soil temperature and humidity in the afternoon, Ds, DMG, SB, V%, MO, H + Al and CTC. It was also noticed that the period of evaluation of the CO2 emission of the soil does not interfere in the obtained data, since the variability in the different periods was similar.