Victor,
If what you mean by "derive an equation for the model to explain the
yield" is that you want to describe how well the model predictions "fit"
your measured yields, you can do this in several ways. You should
definitely read Volume 3, Chapter 2 of the DSSAT4 documentation (in
DSSAT 4 click on the word "Reference" near the bottom of the vertical
toolbar on the left edge of your screen, then click on the icon labeled
"DSSAT4 vol3". This opens an Adobe acrobat file - go to Chapter 2
"Short Description of the Model Statistical Package and Weather Analogue
Program". This describes several methods of statistically measuring the
"fit" of your predictions - it is short but very helpful.
If you are using the cereal model (ex. Maize) you can use the STATS
program mentioned in the article above. It comes with DSSAT4 (this is
an optional program and you may or may not have installed when you
installed DSSAT4 - if not, put your DSSAT4 install disk in your computer
and search through the install menu and install just the STATS
program). To use the program, follow the instructions in the chapter
that I mentioned above.
If you want to compare individual runs (results from 1 simulation
compared to measured values), you can run your simulations and when it
has completed, click on the icon labeled "Graphical Display" found under
"TOOLS" in the left side toolbar in DSSAT4. This loads a program called
"GBUILD" that will graph the results of your simulation runs. If you
have put your measured (observed) data in a "T"-file (ex. .MZT for
maize), GBUILD will plot both the simulated and measured data together.
Next to the graph, you will see a button labeled "Statistic". This will
give you the observed and simulated means, their ratio, standard
deviations for both the observed and simulated data, difference between
observed and simulated results, the Root Mean Square Error (RMSE) of
your predictions, and Willmott's index of agreement (d-index). To build
a "T"-file, look in the DSSAT 4 subdirectory for the plant species that
you are modeling. Look at some of the "T"- files there and try to copy
their format (use a text editing program like Windows Notepad to view
the old files). There is an icon labeled "Experimental Data" in the
DSSAT left-side toolbar (listed under Tools) that will help you build
the T-file and make sure your data are in the correct columns. The
"T"-file must have the same name as the experiment (X) file of your
simulation but the last letter of the names extension should be "T"
instead of "X". For example, the experiment file would be named
UFGA9701.MZX and the "T"-file will be named UFGA9701.MZT.
If you want statistics for the overall fit of all of your runs and don't
want to use the STATS program, you can copy the predicted yields from
the PlantGro.OUT file (found in the DSSAT subdirectory that corresponds
to the plant species that you are modeling) into a spreadsheet.
Alternatively, plot all of your simulations in GBUILD and use the
"Export to Excel" button to dump only the growth data to an Excel
spreadsheet (you must have Excel on your computer for this to work). If
you don't have Excel you can use the "Export to Text file" button to
create a text file with just the growth data - then paste that into a
spreadsheet. Place the observed data in one column and the predicted
yield in the next column. Create a graph in your spreadsheet program
with the Observed data on the X-axis and the predicted data on the
Y-axis. This is your 1:1 comparison. You can now run a linear
regression on this data to fit a line to this plot. You will end up
with an equation with a slope and Y-intercept describing the fit and an
r2 value that indicates the amount of variation that the model can
account for. You can also go ahead and calculate standard deviations
and a d-index for all of your simulations (see the article mentioned
above and/or a good statistics textbook for the equations).
Once the data is in a spreadsheet, you can easily paste it into a
statistics program if you want more involved statistics.
Hope this helps,
Stu
Victor M. Rodríguez M. wrote:
> Hi, everyone on the list,
> I've a couple of questions, but firstable i would like to thank to all the ppl who help me on my prior question related with some troubles on running DSSAT, G. Hoogenboom, J.I Lizaso and JW White.
> 1.- Once we finished the simulation process, is it possible to derive an equation for the model to explain the yield ?
> 2.- Is there any way to calculate the radiation value using only Tmx, Tmin and precipitation daily data ?
>
> T, i, A,
>
> MC Victor M Rodríguez Moreno
> Sensores Remotos y Sistemas de Información Geográfica
> INIFAP-CIRNOC-CEPAB
> Laboratorio Nacional de Modelaje y Sensores Remotos
> Campo Experimental Pabellón
> km 32.5 Carretera Ags-Zac CP 20660
> tel/fax (465) 958 01 61 y 67, Ext 132
> [log in to unmask]
> ----- Original Message -----
> From: Victor M. Rodríguez M.
> To: [log in to unmask]
> Sent: Thursday, February 02, 2006 10:48 AM
> Subject: simulation troubles
>
>
> Hello everyone on the list i am almost newly on DSSAT son hopefully this won´t be a dump question. I m on a run to simulate maize yield and I followed all the route of DSSAT to load experimental data, crop management data, weather data and soil data. I established the simulation options to generate yield but my problem is that when I select the options in "weather" and "models" menu, on the window to select the experiments and simulation outputs, only appears weather.out. So in consequence the results are very unhopeful for me because i cannot see any yield simulation. What could be wrong ? any idea ? Regards
>
>
> MC Victor M Rodríguez Moreno
> Sensores Remotos y Sistemas de Información Geográfica
> INIFAP-CIRNOC-CEPAB
> Laboratorio Nacional de Modelaje y Sensores Remotos
> Campo Experimental Pabellón
> km 32.5 Carretera Ags-Zac CP 20660
> tel/fax (465) 958 01 61 y 67, Ext 132
> [log in to unmask]
>
>
>
--
The Ag Guy, LLC
Stuart Rymph, Ph.D., CCA - Wisconsin and Northeast Regions
6063 Beckman Rd.
Mazomanie, WI 53560
Phone (608) 469-8538
Phone/Fax (608) 795-0021
E-Mail: [log in to unmask]
|
