I am using HYDRUS 2D to create a model for simulating moisture conditions in soil profile by varying water table depths. Which statistical method is apt for validating my model?
Krishna,
Here is some directions (Link)
https://www.researchgate.net/post/How_to_compare_Hydrological_Model_Results
Hello again,
The value of RMSE in perfect modelling is Zero, but the accepted value of RMSE is data-dependent and arguable. Using RMSE is recommended to compare the overall accuracy of results across several models.
Also R2 has some shortcomings:
https://www.researchgate.net/post/Do_you_know_any_good_papers_on_R_Squared_for_physiological_measures
If you want to compare a single components of the results, use relative error in percent (e.g. Peak value of observed and simulated hydrograph).
If your target is comparison of continuous Obs/Sim values, using Nash-Sutcliffe efficiency criteria and its modified versions could be more appropriate.
Having a look on the following papers may give you some insights on the range of acceptable values of efficiency measures.
http://link.springer.com/chapter/10.1007%2F978-3-642-19059-9_3
Dear Krishna,
A good measure to assess the performance of a model against the measured data is the Nash–Sutcliffe efficiency coefficient.
For this, and several other measures of performance you can refer to:
Moriasi DN, Arnold JG, Van Liew MW, et al. (2007) Model Evaluation Guidelines for Systematic Quantification of Accuracy in Watershed Simulations. Trans ASABE 50:885–900. doi: 10.13031/2013.23153
Dear Sir
I think that you can use from Nash–Sutcliffe efficiency coefficient too.
I recommend using several indices of agreement are implemented in HydroGof R, look on:
cran.r-project.org/web/packages/hydroGOF/vignettes/hydroGOF_Vignette.pdf
cran.r-project.org/web/packages/hydroGOF/hydroGOF.pdf
Best regards
@Luis,
Nice suggestion!
@Krishna, Also You can try this:
Online calculation of statistical measures to compare observed and simulated time series data:
www.hydrotest.org.uk
I am assuming that you are validating your Hydrus model against some known soil moisture? The key to successful matching of known and modeled moisture will be the hydraulic conductivity function and parameters. Hydrus has inverse modeling functionality that will allow you to estimate this based on the data you have. It will use a least squares approach in the objective when fitting these two datasets.
Nash-Sutcliffe is one of the good statistical method to measure the goodness of calibration.
In my opinion, RMSE is a suitable criterion for validating the model.
Dear Krishna P Kaja,
I'm using Nash-Sutcliffe efficiency, relative MAE (RMAE) and MAE criteria for calibrating and validating your model.
Regards,
Mr. Van Van THAN
- Badges
- Science topic