Dynamics of streamflow predictability and memory in response to hydrological extremes: insights from the Bandar Abbas watershed.

Abstract

Hydrological extremes, worsened by climate change, disrupt river flow, threatening water resources in arid regions. This study analyzed data from 17 hydrometric stations (1981-2017) across two periods to assess drought impacts on flow behavior in the Bandar Abbas watershed, Iran. Using the Stream Drought Index, noise variance, Lyapunov exponent (LE), Hurst coefficient (H), approximate entropy (ApEn), and the Mann-Kendall trend test, we identified significant changes. Drought severity increased, with noise variance in drought patterns rising by up to 3475% in central zones and dropping by 70% elsewhere, signaling more frequent and intense dry spells. River flow declined at 12 stations, with the Mann-Kendall test confirming negative trends, reducing water availability by up to 150% in volume at northern sites. LE decreased by 1165% across most stations, indicating drought lowered flow sensitivity to initial conditions, while H fell by 50%, weakening long-term flow memory. Meanwhile, ApEn surged by 354%, reflecting increased randomness and reduced predictability, particularly in northern areas. These shifts strain water availability for ecosystems and agriculture, disrupt aquatic habitats, and challenge management strategies reliant on stable flow. This multi-tool approach, novel in this context, merges chaos, memory, and randomness analyses to clarify drought's effects. Focused on Bandar Abbas, the findings offer insights for arid regions globally, providing a framework for adaptive water management to address scarcity and unpredictability in river systems under climate stress.