Dear experts in hydrological modeling, I am new to the HEC HMS model and would like to apply it to simulate the streamflow discharge on my delineated catchment with its 5 sub-basins. I have daily rainfall data and few daily observed discharge data.
I had a problem with how to select the events for calibration and validation. I understand how to calibrate but I don't understand how validation is done? To my eyes in different articles, it seems like the way of calibrating (using the Nash Sutcliffe efficiency, R square, relative error, among others) is the same way of validating (which I think is not true). How is validation done?
Moreover, I have 5 sub-basins. Does it mean that I have to have observed discharge data for 5 stations (for each sub-basin)? what if only one station is available for the whole catchment? (means the observed flow data will be the same for each sub-basins? does it make sense?) I am getting confused about that.
Will I have to produce simulated flow for each sub-basin per event? or only simulated flow for the selected events? what if I want to calculate the simulated flow for the entire catchment?
Thank you very much for your helpful feedback
Dear Richard Mind'je
The Nash Sutcliffe efficiency, R square, relative error, among others are not the method of calibration. Those are just statistical indices which represent the accuracy between simulated and observed values.
You can select 80% of your data for modeling and calibration, and 20% data for your validation (Some modeler also uses 60%-40%, depends on individuals).
You have to change your modeling parameters (this can be roughness coefficient, infiltration index, soil index, runoff coefficient) to simulates the results which gives better match to observed results (check this by calculating statistical indices). When you get the satisfied results, we will call it as calibrated model.
Now you have calibrated all the modeling parameters, don't change those parameter and run the model for the remaining 20% data and calculate the statistical indices for simulated and observed results. If it matches, then your model will be validated.
Nikunj Mangukiya I would like to have some more clarification. "from the values of modeling parameters (before and after optimization), can one know or discover that the used data (precipitation and discharge) are erroneous or incorrect?
Richard Mind'je
As per my knowledge, there is no such option or method which can give you insight about error in data.
Nikunj Mangukiya Dear Sir, Greetings again
Thank you very much for your previous responses that were helpful. However, there is another question passing into my head. Leaving aside the volume produced by the model, is there anyhow we can calculate the volume of the produced peak discharge? or any idea of how we can manually calculate the volume?
Thank you again
Richard Mind'je Yes, there are different empirical methods to determine peak discharge for particular basin. Then you can calculate the volume from the empirical hydrograph. you can refer to following links:
https://www.yourarticlelibrary.com/water/peak-flow/top-5-methods-of-estimation-of-peak-flow/60595
https://www.sciencedirect.com/science/article/abs/pii/S0022169412005008
http://onlinepubs.trb.org/Onlinepubs/trr/1983/922/922-005.pdf
https://www.hindawi.com/journals/geography/2013/214140/
https://ascelibrary.org/doi/10.1061/%28ASCE%29HE.1943-5584.0000114
- Badges
- Science topic
- Similar topics
- Computer Communications (Networks)
- Simulators