Three-Day Delay in Daily Streamflow Simulation Related to Spatial-Object Routing in SWAT+

[北科大-吳邦源]

Hello SWAT+ community,

My daily streamflow simulation shows that the simulated peak flows occur approximately three days later than the observed peaks. During calibration,

I adjusted parameters such as “surlag” and “aquifer_delay”, but neither parameter had a significant effect on the peak timing. However, after modifying the spatial-object connections so that outflow from each landscape unit (LSU) and aquifer routed directly to the next downstream channel, extract the “flow-in” data of the same reach to compare observed data. The results show that simulated peak timing became closer to the observations, and the

Nash–Sutcliffe efficiency coefficient (NSE) improved significantly. Since all other inputs and parameters remained unchanged, could this indicate that routing path among spatial objects in SWAT+ accumulates a time delay when using a daily simulation time step?

My model information is as follows:

• SWAT+ Editor version: [3.1.4]
• SWAT+ executable revision: [61.0.2]
• Simulation time step: daily
• Watershed area: approximately 303 km²
• Calibration variable: daily reservoir inflow
• Original NSE: [-0.05,0.06]
• NSE after modifying the routing connections: [0.22,0.3]
• Surface runoff method: Curve Number (CN) method

I would appreciate any suggestions regarding whether this behavior is expected or whether the routing configuration should be revised.

Thank you.