Efficiency

Tags: not added yet

DSL Element - `efficiency`

Model efficiencies are commonly used to quantify the prediction performance of a simulation model by computing some aggregate based on observed and simulated values of the same model property. Multiple efficiencies can be computed at once. Just combine those by using the '+' operator as shown. The list of available model efficiencies is shown below at the bottom of this page. The simulation below shows the use of the efficiency element in a simulation.

sim(name:"Efcarson") {
   // define the model
   model(classname:"model.PrmsDdJh") {
   // ... model sub-elements here
   }
   efficiency(obs:"runoff[0]", sim:"basin_cfs", methods:NS+NS2+ABSDIF+TRMSE)
}

Executing the simulation above produces additional table output for the requested efficiencies, which as an example we have rendered as follows:

Efficiencies	ns1	ns2	absdif	trmse
runoff/basin_cfs	0.66512	0.82971	764.30044	2.44043

Element `efficiency`

Name
`efficiency` - Model efficiency computation

Properties	Description	Type	Required
`obs`	an output field that provides observed values	string	yes
`sim`	an output field that provides simulated values	string	yes
`precip`	precipitation values	string	yes only for `ROCE`, ignored otherwise
`method`	efficiency method(s) to compute	KEYs (can be combined)	yes
`file`	the output file	string	no if missing output goes to the console, otherwise to the specified file located in the output folder

Parent
`sim{}`

Notes

Multiple keys can be combined using the + operator, the output will be a combined table.
Multiple efficiencies can use the same file for output. They get appended.

Several model efficiencies are available, as shown in the following table

Efficiencies (KEY)	Description
`ABSDIF`	Absolute difference
`LOGABSDIF`	Log of the absolute difference
`NS`	Nash-Sutcliffe
`LOGNS`	Log of Nash-Sutcliffe
`LOGNS2`	Log of Nash-Sutcliffe (Pow 2)
`IOA`	Index of Agreement
`IOA2`	Index of Agreement (Pow 2)
`R2`	Goodness of fit
`GRAD`	Linear Regression Gradient
`WR2`	Weighted Correlation Coefficient
`DSGRAD`	Double Sum Analysis Gradient
`AVE`	Absolute Volume Error
`RMSE`	Root Mean Square Error
`PBIAS`	Percent BIAS
`PMCC`	Pearson product-moment correlation coefficient
`TRMSE`	Transformed Root Mean Square Error
`ROCE`	Runoff Coefficient Error