Model Evaluation with FluEgg
From Moriasi et al. (2007):
Model evaluation refers to the applicable steps of sensitivity analysis, calibration, validation, uncertainty analysis, and application.
We recommend that both graphical techniques and quantitative statistics be used in model evaluation.
Provides an initial overview of the model performance against observed data and are essential to appropriate model evaluation (Moriasi et al., 2007)
FluEgg > Load river input File > HEC-RAS Evaluation
- Click Load HEC-RAS project file... button and navigate to your Hec-Ras project (*.prj)
- Click Load Observed Data button and navigate to your tab-delimited file containing observed data
- Select the variable of interest: Flow or Water Depth
- Select the River, Reach, River Station and Plan for comparison with observed data
- Click Import TS button
- Click Plot button
You should get a hydrograph with HEC-RAS outputs, observations, and statistical measures.
The Observed data must be in a tab-delimited (->) ASCII file as follows:
A -> -> MARBLEHEAD
B -> -> 29HRS
C -> -> STAGE
E
F -> -> NOAA
Units -> -> meters
Type -> -> INST-VAL
1 -> 04Jun2015 -> 0000 174.333
2 -> 04Jun2015 -> 0006 174.325
3 -> 04Jun2015 -> 0012 174.328
4 -> 04Jun2015 -> 0018 174.336
5 -> 04Jun2015 -> 0024 174.331
...
An easy way to obtain a formatted file is exporting a DSS file using the Tabulate in MS Excel in HEC-DSSVue 2.0.1. This will automatically format the file for you. In MS Excel, save the file as Tab-delimited.
The figure below is an example of a hydrograph of HEC-RAS results and observed data for the Sandusky River, OH. It includes statistical measures (PBias and Nash-Sutcliffe).
PBIAS is a measured of the average tendency of the model to under-, or over- estimate the observations. It is defined as the percent ratio of the cumulative absolute error to the sum of the observations.
PBIAS = \left(\frac{\sum_{i=1}^{n} (Y_i^{obs} - Y_i^{sim})*(100)} {\sum_{i=1}^n(Y_i^{obs}} \rigth)
A value of PBIAS = 0.0, indicates an overall perfect match of the simulated data with the observations. Low-magnitude PBIAS are indicative of accurate model estimates (Moriasi et al., 2007). Negative values indicate an overall overestimation by the model, while positive values reflect an underestimation on the simulation results.
NSE is the ratio of the error variance to the observations variance. Perfect match between simulated and observed data results in NSE = 1 (maximum). Values between 0 and 1 are acceptable. Negative values correspond to unacceptable model-performance, since the mean of the observed values is a better predictor than the model (Moriasi et al., 2007).
The NSE is computed as follows
NSE = 1 - \left( \frac{\sum_{i=1}^{n}(Y^{obs}_i - Y^{sim}i)^2}{\sum{i=1}^{n}(Y^{obs}_i - Y^{mean}_i)^2} \right)
Moriasi, D. N., Arnold, J. G., Van Liew, M. W., Binger, R. L., Harmel, R. D., & Veith, T. L. (2007). Model evaluation guidelines for systematic quantification of accuracy in watershed simulations. Transactions of the ASABE, 50(3), 885–900. http://doi.org/10.13031/2013.23153