Hydrological seasons in the Brazilian semi-arid region: database organisation, seasons identification, and modelling of river intermittence

Abstract

The rainy season is extremely important in the hydrology of the Brazilian semi-arid region, due to biological and economic activities, as a series of hydrological processes can be better observed in this period. Due to the high variability of precipitation in the region, the study of seasonality is an essential factor for hydrology. Intermittency is predominant in most rivers in the Brazilian semi-arid region. Hence, the importance of studying water distribution along the river, especially in periods without continuous flow, is evident. The objective of this work is to evaluate the temporal variability of hydrological processes, considering the dynamics of this system in the Brazilian semi-arid region. The study area corresponds to the Benguê reservoir basin (1000 km²) and the Aiuaba Experimental Basin – AEB (12 km²), nested into the Benguê basin. The thesis is composed of three thematic chapters: i) Organisation and structuring of the AEB database; ii) Identification of hydrological stations and analysis of two decades of monitoring; and iii) Modelling river intermittency with drone data and random forest models. The first chapter deals with the compilation and systematisation of data acquired over the last 20 years at AEB (2003 to 2022). Based on data from the database, a method was developed to delimit hydrological seasons in the study area with data series of precipitation, soil moisture and spectral response of vegetation. In Chapter III, there is the mapping and modelling of the intermittency of the main river (Rio Umbuzeiro) of the Benguê basin. The organisation and cataloguing of data acquired over the last 20 years at AEB was important for a better hydrological understanding of the region, contributing enormously to current and future work in the region and facilitating database maintenance and use. The proposal and application of the new method for identifying hydrological stations allowed the characterisation of seasons in the Brazilian semi-arid region using criteria based on precipitation, soil moisture and vegetation. It was concluded that the application of the method allowed the study of trends related to seasons, making it possible to identify the trend towards an increase in the rainy-dry transition season. Modelling river intermittency was possible with an accuracy of around 80% for models developed in the work. The application of the model to the entire river allowed an assessment of the spatial and temporal distribution of river intermittency, concluding that the model using Sentinel images with the Modified Normalised Difference Water Index (MNDWI) had the best performance in extrapolating the model to the entire river.