Multivariate time-series modeling and forecasting of groundwater level fluctuation in the Erbil basin, North of Iraq

Abstract Groundwater constitutes a critical supply for domestic, agricultural, and industrial needs. Understanding the dynamic interplay between rainfall, river discharge, and groundwater levels remains challenging due to the complexity of hydrogeological structures and variable recharge conditions. This study investigates the temporal relationships among rainfall, river stage-discharge, and groundwater levels in the Erbil Basin, North Iraq. A multivariate time series framework was employed for the data series spanning 2004–2022, combining wavelet coherence analysis to identify dominant periodicities and temporal correlations, with a vector autoregression (VAR) model to forecast future groundwater fluctuations. The wavelet analysis revealed strong monthly-scale variability during the wet season and a pronounced rainfall-groundwater coherence with a dominant 1-month periodicity, particularly after 2008. Spatial differences in coherence strength across monitoring wells indicate regionally consistent hydrogeological conditions, with stronger coupling observed in wells situated closer to the Lesser Zab River. The impulse response function (IRF) analysis additionally demonstrated positive groundwater responses within 2–6 months following rainfall events, consistent with regional recharge patterns. However, persistent groundwater declines identified in the VAR forecasting model reflect the combined effects of reduced recharge and continuous abstraction pressures. The integrated approach provides a comprehensive assessment of hydroclimatic interactions and groundwater dynamics in the Erbil regional aquifer. The findings emphasize the need for adaptive groundwater management strategies and demonstrate the potential of combining wavelet and VAR models for sustainable water resource planning in data-limited, semi-arid environments.