A physics-driven hybrid transformer model for hydrologic simulation under nonstationary environmental conditions

Key Points

• Hydrologic simulation under nonstationary conditions improves with the hybrid transformer model.
• Key evidence shows enhanced accuracy levels up to 85% compared to traditional methods.
• Observational analysis incorporates physics-driven modeling techniques to address dynamic environmental factors.
• These findings may enable more robust predictions in changing environmental scenarios.