Hi,
The simulated peak in HEC HMS model is one day after the observed peak.
The method which i am using:
1. Muskingum method for routing
2. SCS curve number for loss
3. SCS unit hydrograph for Transform method
4. Constant monthly method for Baseflow
I have done changes in certain parameters just to match the peak time so, the NSE is 0.857 but percent bias is 50%.
As i am unable to figure out how to reduce this 50% percent bias.
2nd question is:
What will be the perfect duration for validation w.r.t calibration in general.
Presenting the plots for the results from calibration will be helpful in analysing the issues in the model.
Calibration and validation time periods vary, ideally it should be an even split with statistically similar data sets. However, researches prefer to use a larger time period for calibration such that the validation results are more preferable.
I'm not experienced with HEC HMS; but how long are your time steps--are time steps one day, or shorter?
Hi Usama--with model time steps of 1 day, it could be that the temporal resolution of the model results have an uncertainty of at least +/- 1 day; depending on details of the model numerical computation mechanics. Perhaps the HEC HMS model documentation details the model computation mechanics. I would guess it may be difficult to set the model up with hourly time steps (may not be worthwhile unless the observed data has sub-daily, preferably hourly, time steps). Before undertaking such a finer time-scale modeling effort, Omesh's recommendation of plotting the observed data together with model results should help you to more reliably diagnose the source of the discrepancies in your model results.
- Badges
- Science topic
- Similar topics
- Filing