Avaliação da co-simulação sequencial gaussiana na geração de campos aleatórios da condutividade hidráulica do solo saturado

Abstract

The dynamics of soil water is a central point for hydrological cycles in watersheds; however, the spatial representation of processes related to soil water movement in mathematical models still represents a major challenge, given the complexity of these interactions. Saturated soil hydraulic conductivity (KSat) is a key input parameter for modeling soil water flow and applicable to irrigation and drainage projects, for modeling runoff, and for containing soil and water loss by erosion. There are many ways to determine KSat, with in situ and laboratory measurement methods. However, these analyzes are expensive and time-consuming, especially at the watershed scale, due to the magnitude of the areas and also the great heterogeneity of the soil and landscape characteristics that control this phenomenon. The present work aims to understand the interactions of soil hydrological attributes and landscape characteristics with KSat from multivariate statistical analysis and, subsequently, apply these attributes/characteristics as secondary variables for spatial modeling of KSat using geostatistical methods, improving the results obtained and reducing the associated uncertainties. This thesis was divided into two complementary but independent original papers. The first paper is entitled "Examining the spatial relationship between saturated soil hydraulic conductivity and co-driven factors at a watershed scale using multivariate and geostatistical methods" and the second paper is entitled "Investigating the spatial uncertainty of saturated soil hydraulic conductivity and co-regionalized variables at the watershed scale using multivariate and Gaussian co-simulation analyses”. The research for the elaboration of both original papers was conducted in the Ellert Creek Watershed, which has about 70 hectares where land use and combine agricultural production areas and native forests. Soil and landscape information was collected at 178 sampling points distributed iat the watershed area from a terrestrial and remote survey. On the first paper, the maps of the two principal components showed a very close relationship with land use and soil textural classes, allowing to state that these variables are systematic for KSat in the study area. For the second paper, among soil hydrological attributes, landscape characteristics and principal components, the results demonstrate that soil density and macroporosity are the most reliable secondary variables to improve the spatial evaluation of KSat by Sequential Gaussian Co-simulation.