SWAT+ Toolbox Calibration: Persistent Peak Overestimation and NSE Stagnation

[Khyber]

Dear colleagues,

I am currently working on calibrating monthly streamflow using the SWAT+ Toolbox (DDS algorithm) and would appreciate any insights from the community.

Model setup:

• SWAT+ with monthly calibration (Channel flow)
• Simulation period: 2006–2014 (2-year warm-up)
• Observed data: 2008–2014
• Objective function: NSE
• Calibration algorithm: DDS (400 iterations)

Current status:

• Initial calibration reached NSE ≈ 0.43
• After refining parameter ranges (to reduce peak overestimation), the calibration became stable but NSE dropped (~ -0.05 to low positive values in early iterations)
• Dotty plots initially showed extreme values (now resolved after resetting runs)

Main issue:

• Simulated peaks are consistently higher than observed values
• Model tends to push parameters toward upper bounds (e.g., AWC, PERCO, BF_MAX), suggesting a need for increased storage/attenuation
• However, constraining ranges too strongly leads to poor model performance

Parameters currently used:
alpha, revap_co, revap_min, bf_max, cn2, canmx, perco, cn3_swf, awc, chk

Observation:
It seems that adjusting soil and groundwater parameters alone is not sufficient to control peak flows effectively. I am considering whether routing/timing parameters (e.g., lat_time or others in SWAT+) should be included in calibration.

---

My questions:

1. Has anyone experienced similar persistent peak overestimation in SWAT+ Toolbox calibration?
2. Which additional parameters (especially routing or lag-related) would you recommend including?
3. Any guidance on balancing parameter ranges without over-constraining the search space?

---

Any suggestions or experiences would be highly appreciated.

Best regards,

[Muhammad Nawaz]

I am also getting an overestimation in the simulated flow due to my rainfall data, which is not accurate. 

I suggest checking your rainfall data.

--
You received this message because you are subscribed to the Google Groups "SWAT+ Toolbox" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sptoolbox+...@googlegroups.com.
To view this discussion, visit https://groups.google.com/d/msgid/sptoolbox/15ee7e1f-604c-472f-b696-44d9294e6d92n%40googlegroups.com.