Série hidrológica não estacionária e os riscos e incertezas nas tomadas de decisões no planejamento dos recursos hídricos

Abstract

Hydric planning aims to guarantee, in quantity and quality, the availability of water for current and future generations. However, due to various environmental changes in land use, in addition to global climate changes, it has become impossible to assume a stationary condition of the hydrological series. Hydrological series, the fundamental principle of water planning, is based on the implicit assumptions of randomness, independence, homogeneity. Due to such modifications, the notion that natural phenomena circulate in a fixed process of uncertainty may no longer exist. That is, the most fundamental basis of predictive statistics may have its bases undermined. Nevertheless, deterministic hydrological models and stationary assumptions are widely applied in projects, plans, instruments, reference flow, without the use of sensitive non-stationary detection methods. In this context, the main objective of this study was to evaluate the uncertainties and risks of non-stationarity in planning the use of hydric resources and to evaluate the distribution of stationary segments of the non-stationary time series. The R software was used to perform statistical tests and, for segmentations, the analysis of waves, justified because it preserves local, non-periodic and multiscale phenomena, appropriate for analyzing irregular, poorly distributed events and potentially non-stationary series at different frequencies. This methodological approach was applied at Station 87440000 - Passo das Canoas of the Gravataí River Watershed. By analyzing the frequency distribution of stationary segments in this non-stationary series, one can verify uncertainties of up to 70% in stationarity for trend and up to 80% in stationarity for level. The permanence frequency curves of the stationary segments for level and for trend showed restriction of up to 99% of the time. In addition, maximum periods of 2 years for trend and 3 years for level were found. Therefore, it was concluded that the use of stationary methods in watersheds with changes in land use and climate have high risks and uncertainties, making its management difficult. In addition, there is a need for statistical methods that are more sensitive to non-stationary hydrological series. The recommended path is the adaptive management conditioned to a decrease in the planning horizon in the watersheds with non-stationary hydrological series.