Modelos geoestatísticos para a gestão sustentável de recursos hídricos em Regiões Cársticas.

Abstract

Karst aquifers are essential water sources, supplying around 7% of the global population. In the State of Bahia, these aquifers cover a vast area and are particularly relevant in the semi-arid region, where they constitute the main water reservoir. Due to their complex morphological evolution, these reservoirs exhibit significant heterogeneity in the distribution of their hydrodynamic parameters. Among these, transmissivity (T) is one of the most important, representing an aquifer's capacity to transmit water through a cross-sectional area. Transmissivity is crucial for understanding groundwater flow and contaminant transport. Its measurement is carried out through pumping tests, which require specific techniques and financial investment, making it difficult to obtain in simpler studies. On the other hand, specific capacity (Sc) is a more easily acquired parameter, usually obtained during well drilling, and can be used as auxiliary data to estimate transmissivity, thus expanding the available dataset for analysis. This thesis aims to fill a gap in the knowledge regarding the spatial variability of hydrodynamic parameters in the Salitre Karst Aquifer, located in the Chapada Diamantina region, in central Bahia. To achieve this, hypotheses were established, such as: which hydrodynamic and geomorphological parameters correlate with transmissivity? Is it possible to spatially estimate T values using geostatistical models? And can uncertainties in T estimates be reduced? To address these questions, two scientific articles were developed and published in the journal Water: the first, titled "Geospatial analysis of transmissivity and uncertainty in a semi-arid karst region," and the second, "Multivariate geostatistics in the spatialization of transmissivity in karst aquifers." The thesis is organized into four chapters: Chapter I provides an introduction, while Chapters II and III correspond to the two articles. Chapter IV presents the conclusions. The research used data from 2,728 pumping tests, provided by the Bahia Water Engineering and Sanitation Company (CERB), the Geological Survey of Brazil (SGB), and the Institute for the Environment and Water Resources (INEMA). These data cover the municipalities of Iraquara, Irecê, and the area near the São Francisco River in the Salitre Karst Aquifer. From this data, linear and spatial modeling of transmissivity (T) and specific capacity (Sc) values were performed, based on analytical models from the literature. In the first article, with 269 Sc values and 51 T values, an empirical linear model for logT based on logSc was established. This model allowed for an expanded dataset for the application of Ordinary Kriging (OK) and Conditional Gaussian Sequential Simulation (SGS). This study focused on the southern portion of the aquifer, around Iraquara, covering the Paraguaçu River basin. In the second article, multivariate geostatistical models were constructed using Ordinary Co-Kriging (COK) and Conditional Gaussian Sequential Co-Simulation (Co-SGS), with 174 primary logT data and 2,728 secondary logSc data. Both studies estimated logT values in unsampled areas and assessed the uncertainties associated with the modeling. COK showed lower uncertainties compared to OK, and Co-SGS proved to be more accurate than SGS. Despite their greater complexity and computational effort, the multivariate geostatistical models stood out for better spatialization of transmissivity and uncertainties in the Salitre Karst Aquifer. The results of this thesis can significantly assist in water resource management by enabling the incorporation of transmissivity values and their uncertainties into numerical groundwater flow models, such as those based on finite elements or finite differences. This will facilitate the estimation of water balance and the study of the impacts of aquifer exploitation or contaminant transport.