Actions

TAYLORTARGETDIAGRAM: Difference between revisions

From BAWiki

mNo edit summary
mNo edit summary
Line 15: Line 15:


|shortdescription=
|shortdescription=
The application taylorTargetDiagram (see also [[TAYLORTARGETDIAGRAM.DAT|taylorTargetDiagram.dat]]) allows to visualize variables for the estimation of model performance (e.g. calculated by [[NCDELTA]]) and list them as table. <br />
The application taylorTargetDiagram (see also [[TAYLORTARGETDIAGRAM.DAT|taylorTargetDiagram.dat]]) allows to visualize variables for the estimation of model performance (e.g. calculated by [[NCDELTA]]) as taylor and target diagrams and list them as table. <br />


This Program is a further development of [[TAYLORDIAGRAM]]. <br />
This Program is a further development of [[TAYLORDIAGRAM]]. <br />
Line 22: Line 22:


<gallery>
<gallery>
Taylor.png |Example picture Taylor diagramm
Taylor.png |Example picture Taylor diagram
Target.png| Example picture Target diagramm
Target.png| Example picture Target diagram
TaylorTarget_Tabelle.png|Example picture of the resulting table  
TaylorTarget_Tabelle.png|Example picture of the resulting table  
</gallery>
</gallery>


In addition to the figures, a table including the following statistical values can be created:
In addition to the figures, a table including the following statistical values can be generated:
* Root Mean Square Error (RMSE)
* Root Mean Square Error (RMSE)
* Taylor Skill 4 (Equation 4 in Taylor (2001))
* Taylor Skill 4 (Equation 4 in Taylor (2001))
Line 59: Line 59:


|methodology=
|methodology=
The necessary statistical characteristics are read with [[MATLAB]] from [[NCDELTA]] result files and visualized. The switch „With_Taylor_Diagram_Data=true“ has to be turned on in [[NCDELTA.DAT]] so that the required statistical values are included in the [[CF-NETCDF.NC|cf-netcdf]]-file written by [[NCDELTA]].  <br />
The necessary statistical characteristics are read with [[MATLAB]] from [[NCDELTA]] result files and are visualized. The switch „With_Taylor_Diagram_Data=true“ has to be set to "true" in [[NCDELTA.DAT]] so that the required statistical values are included in the [[CF-NETCDF.NC|cf-netcdf]]-file written by [[NCDELTA]].  <br />
The visualization of model accuracy in a taylor diagram is done according to Taylor (2001).<br />
The visualization of model accuracy in the taylor diagram is done according to Taylor (2001).<br />
The visualization of model accuracy in a target diagram is done according to Jolliff et al. (2009).
The visualization of model accuracy in the target diagram is done according to Jolliff et al. (2009).
<br />
<br />



Revision as of 09:34, 26 May 2020

Basic Information

Name of Program

taylorTargetDiagram

Version-Date

May 2019

Description-Date

May 2020

Catchwords

Taylor-Diagram
Target-Diagram
model accuracy
comparison of model and measurement
CF NetCDF format
MATLAB

Acknowledgement: This project took advantage of netCDF software developed by UCAR/Unidata (www.unidata.ucar.edu/software/netcdf/).

Short Description of Functionality

The application taylorTargetDiagram (see also taylorTargetDiagram.dat) allows to visualize variables for the estimation of model performance (e.g. calculated by NCDELTA) as taylor and target diagrams and list them as table.

This Program is a further development of TAYLORDIAGRAM.

In the Taylor-Diagram standard deviation, correlation as well as pattern root mean square error (PRMSE) between reference and variant (e.g. model and observation) are graphically displayed. The target diagram shows the centered root-mean-square deviation and the bias. The combined presentation of the statistical characteristics of different physical quantities and locations in one diagram is possible. The statistical characteristics at different locations can also be combined to a mean value. As reference and variant values for the calculations of differences and statistcs results of different simulations or analyses as well as comparisons between measurements and model results may be used.

In addition to the figures, a table including the following statistical values can be generated:

  • Root Mean Square Error (RMSE)
  • Taylor Skill 4 (Equation 4 in Taylor (2001))
  • Taylor Skill 5 (Equation 5 in Taylor (2001))
  • Centered Root-Mean-Square-Difference
  • Bias (deviation of the mean values)
  • Mean values of reference and variant parameters
  • Standard deviation of reference and variant parameters
  • Correlation between reference and variant parameters
  • Attributes of the locations of the parameters compared in NCDELTA (cartesian and geographic coordinates, depth, location name)
  • File name of the output file of NCDELTA
  • Number of data points used for the comparison


Example Applications

Input-Files

  1. general input information (file type taylorTargetDiagram.dat);
  2. NCDELTA-Output files format (cf-netcdf.nc).

Output-Files

  1. Taylor-Diagram as bitmap (*.JPG,*.PNG), as a vector graphic (*.PDF,*.EPS) or as a MATLAB figure file (*.FIG).
  2. Target-Diagram as bitmap (*.JPG,*.PNG), as a vector graphic (*.PDF,*.EPS) or as a MATLAB figure file (*.FIG).
  3. (optional) Table of the model performance quantities (*.CSV,*.DAT)
  4. (optional) Text file with meta-information (*.TXT).

Methodology

The necessary statistical characteristics are read with MATLAB from NCDELTA result files and are visualized. The switch „With_Taylor_Diagram_Data=true“ has to be set to "true" in NCDELTA.DAT so that the required statistical values are included in the cf-netcdf-file written by NCDELTA.
The visualization of model accuracy in the taylor diagram is done according to Taylor (2001).
The visualization of model accuracy in the target diagram is done according to Jolliff et al. (2009).

Program(s) to run before this Program

NCDELTA

Program(s) to run after this Program

MATLABFIGVIEWER, Excel

Additional Information

Language

MATLAB

Additional software

Linux: MATLAB Compiler Runtime MCR R2015a in directory /usr/local/MATLAB/R2015a

Original Version

U. Schiller, J. Benndorf, C. Rasquin

Maintenance

U. Schiller, J. Benndorf, C. Rasquin

Documentation/Literature

Taylor, Karl E. (2001) "Summarizing multiple aspects of model performance in a single diagram". Journal of Geophysical Research, Vol 106, No. D7, April 16, 2001, Pages 7183 - 7192.
Jolliff et al. (2009) “Summary diagrams for coupled hydrodynamic-ecosystem model skill as-sessment”. Journal of Marine Systems. Vol 76, Issues 1–2, 20 February 2009, Pages 64-82, ISSN 0924-7963
Siehe auch unter MATLAB.

  • Example files:
    • Example files can be found in $PROGHOME/examples/taylorTargetDiagram/
    • A presentation with further examples and information can be found in $PROGHOME/examples/taylorTargetDiagram/_doc

back to Program Descriptions


Overview