Extremos de precipitação e disponibilidade hídrica na América do Sul sob mudanças climáticas.

Abstract

This thesis investigates the future reorganization of precipitation extremes and climatic water availability (CWA) in South America and tests the hypothesis that this reorganization occurs nonlinearly across different levels of warming, exhibits spatial heterogeneity, and cannot be accurately assessed without accounting for model biases, intermodel spread, and the limitations of analyzing precipitation alone. To test this hypothesis, the thesis combines bias correction and downscaling of Coupled Model Intercomparison Project Phase 6 (CMIP6) simulations using Quantile Delta Mapping (QDM); the comparison of ensemble methods and probabilistic projections of extremes under the SSP2-4.5 and SSP5-8.5 scenarios; and the representation of CWA through the nonparametric SNIPE index, whose relationship with historical teleconnections is quantified using multiple linear regression. QDM reduced spatial biases and improved the representation of duration indices but increased interannual errors in accumulated totals and maximum intensities. The independence-weighted ensemble performed best among the methods evaluated, frequently outperforming individually corrected models. The projections reveal nonlinear responses between scenarios, particularly in the Northeast, and indicate stronger drying and a greater concentration of rainfall in short, intense events in the Amazon and Northeast; more intense rainfall early in the year and greater spring aridity in the Northwest; a delayed rainy season in the South American Monsoon System; wetter summers and autumns and a drier spring in the Southeast; wetter winters and drier warm seasons in the South; and persistent drying accompanied by a decline in winter rainfall in the Southwest. Reductions in CWA are projected for the Amazon, the Northeast, and the southern coast of Chile, whereas increases are projected for the La Plata Basin and southern Brazil. In the historical analysis, the regression identified three regimes of teleconnection influence on CWA: a predominance of the El Niño–Southern Oscillation (ENSO) in the equatorial sector, with a comparable contribution from the Atlantic in some areas; a limited contribution from ENSO, when analyzed separately, at locations in the Northeast and Central Brazil, where low-frequency variability in the Atlantic and Pacific exerts a stronger influence; and a small fraction of variance explained by the selected teleconnections in the Southern Cone, consistent with a greater role of internal variability and transient synoptic systems. Taken together, the results support the hypothesis. The main contribution of this thesis is to show that future hydroclimatic risk cannot be reduced to a general pattern of drying or wetting, but instead combines regional losses of rainfall regularity, a greater concentration of precipitation in intense events, and changes in CWA, while also demonstrating that the statistical adjustments improve the accuracy of projected values for most precipitation-extreme indices.