Study region: Iran. Study focus: Flash droughts are intensifying under climate variability and warming, yet their dominant controls remain insufficiently understood across climatically heterogeneous regions such as Iran. This limits the development of robust and interpretable drought monitoring and early-warning systems. To address this gap, we conducted a climate-dependent assessment of flash drought intensity across 110 major river basins in Iran, classified into four aridity-based climatic zones (very dry, dry, semidry, and humid). Flash droughts were identified from pentad-scale soil moisture percentile dynamics, and continuous drought intensity was modeled using three complementary frameworks: CatBoost coupled with SHAP, Long Short-Term Memory (LSTM) integrated with DeepSHAP (Integrated Gradients), and a pattern mining-based ensemble (PME) implemented with Random Forest. A comprehensive set of hydro-meteorological variables, including precipitation, multi-layer soil moisture, temperature extremes, relative humidity, wind speed, solar radiation, and reference evapotranspiration, was analyzed over the period 1979-2023. Model evaluation was conducted primarily against SPEI-3 and supplemented by SPEI-1 to examine robustness across benchmark timescales. New hydrological insights for the region: The results reveal that flash drought intensity across Iran is governed by systematic, climate-dependent transitions between soil moisture storage limitation and atmospheric evaporative forcing. In very dry regions, which cover approximately 70% of the study area, drought intensification is predominantly controlled by persistent depletion of multi-layer soil moisture, reflecting a strongly storage-limited hydrological regime with minimal buffering capacity against atmospheric stress. In dry regions, drought dynamics exhibit a coupled moisture-energy control, where precipitation deficits and enhanced evaporative demand jointly regulate drought development and persistence. Semidry regions show the highest sensitivity to short-term atmospheric anomalies, where rapid drought intensification is triggered by fluctuations in temperature and relative humidity, indicating a transitional land-atmosphere coupling regime highly responsive to sub-seasonal variability. In contrast, humid regions are primarily governed by atmosphere-driven processes, in which elevated temperature and evaporative demand dominate drought intensification despite relatively higher baseline soil moisture availability, reflecting an event-driven drought regime controlled by episodic atmospheric forcing. Across all climatic zones, consistent hydroclimatic responses derived from both SPEI-3 and SPEI-1 indicate a robust and systematic shift from storage-controlled (memory-driven) drought dynamics in arid environments toward atmosphere-controlled (forcing-driven) drought behavior in humid climates. This transition highlights the fundamental role of land-atmosphere interactions in modulating drought intensity across hydroclimatic gradients. These findings provide a process-based basis for improving sub-seasonal drought predictability by explicitly linking soil moisture memory effects with atmospheric forcing, thereby supporting the development of more effective early-warning systems and climate-adaptive water and agricultural management strategies in regions with strong climatic contrasts.