Author(s): Xin Li; Yibin Zhou

Linked Author(s): Li Xin

Keywords: Weather generator spatial correlation inter-variable correlation inter-annual variability

Abstract: Developing stochastic hydrometeorological fields with spatiotemporal correlations and physical coherence is essential for hydrological simulations. This study proposes a multisite, multivariate weather generator that integrates a coupled multivariate first-order autoregressive (MAR1) model, a first-order Markov chain, and a K-nearest neighbors (KNN) model. The generator simulates daily precipitation, maximum temperatures, and minimum temperatures across 12 secondary water resource divisions in the Yangtze River Basin, capturing spatiotemporal dependencies, inter-variable correlations, and low-frequency interannual oscillations. Evaluation metrics, including statistical characteristics, correlation features, and interannual variability, demonstrate the model’s ability to replicate observed meteorological patterns. However, the generator underestimates maximum drought and wet period durations and first-order autocorrelations for temperatures at some stations. These findings contribute to advancing distributed stochastic hydrological simulations.

DOI:

Year: 2025