Guidance Needed on SWAT-CUP Sensitivity Analysis: 95PPU Not Capturing Peak Flows

[Yalelet Fikrie]

Hello,

I hope this message finds you well and I am a new user of SWAT-CUP currently working on the sensitivity analysis for my SWAT model. I have run into an issue with my initial results and would be very grateful for your guidance.

Here is a summary of my model setup:

· Model Simulation Period: 2005 - 2021

· Warm-up Period: 3 years (2005-2007)

· Observation Period for Analysis: 2015 - 2018

· Current SWAT-CUP Phase: Sensitivity Analysis (using SUFI-2)

The Problem I'm Facing:

After running my first sensitivity analysis,I've found that the 95% prediction uncertainty (95PPU) band is very narrow and fails to capture the observed peak flows in my data. The attached plot clearly shows the highest flow events lying completely outside the simulated uncertainty range. The summary statistics (p-factor ~0.48, r-factor ~0.58, NSE ~0.02) further confirm the model's poor performance in simulating these peaks.

My Understanding & Concern:

As I am in the sensitivity analysis phase,my goal is to see how the model outputs respond to parameter changes. The narrow 95PPU that misses the peaks suggests that my initial parameter ranges might be too restrictive, preventing the model from exploring a solution space that can generate high enough flows.

I would sincerely appreciate your advice on how to proceed. Specifically:

1. Based on the parameters I'm testing, which ones are most critical for controlling peak flows and surface runoff that I should ensure have sufficiently wide and physically plausible ranges?

2. Is it standard practice at this stage to significantly widen the parameter ranges and re-run the sensitivity analysis to ensure the 95PPU can capture the observed data spread?

3. Should I be focusing on a specific subset of parameters for peak flow in my initial analysis?

Any advice or pointers to helpful resources would be immensely helpful as I navigate this learning process.

Thank you so much for your time and consideration.

Best regards,