A frequency-domain nonstationary multi-site rainfall generator for use in hydrological impact assessment

Growing concerns about the hydrological impacts of climate variability and climate change suggest an imperativeness to generate plausible climate scenarios suitable for the vulnerability assessment studies. A frequency-domain nonstationary framework for multi-site rainfall generation is proposed for decision-centric hydrological impact assessments. The framework has three main components: (1) a spatiotemporal rainfall field, described as spatial modes and their corresponding temporal evolution, based on empirical orthogonal function analysis (EOFA); (2) the time series of these spatial modes, decomposed into intrinsic mode functions (IMFs) with characteristic frequencies (periods) using the Hilbert-Huang transform (HHT); and (3) Stochastic simulation (SS), achieved by assigning random phases to the noise IMFs in combination with adjustments both to the residual series and to the signal IMFs. A synthetic WA function is first used to illustrate the power of the EHS (EOFA + HHT + SS) rainfall generator to detect and extract signals (e.g., nonstationary oscillation and trend component) from noisy data. A real application of the EHS model is then presented for the Xiang River basin to demonstrate its ability (reproducibility and adaptivity). The results showed that the EHS rainfall generator has sufficient capacity in reproducing the original spatiotemporal structure, such as the spatial correlation and low-frequency variability. Meanwhile, the EHS model exhibits advantages in terms of perturbing the distribution characteristics of rainfall and altering their behavior according to the intrinsic spatial patterns. These features give the EHS model high feasibility to act as a scenario generator for generating a wide range of possible rainfall scenarios reflecting different aspects of climate variability and climate change, and hence bolster the hydrological impact analysis in the climate change context.