Growth Model Users Group
 Meeting Follow-Up

If you have comments, rejoinders, and/or additions to the meetings or these items, please send them to us and we will post them.

February 11, 2008 Meeting

While noodling around with R and on CRAN I stumbled on an R-Wiki page with high quality graphics tips for R:  I have been using the pdf graphics device for a while but now that I found that there are postscript ( postscript() ) and windows metafile ( win.metafile() ) devices too. Why look beyond R?

March 25, 2006 Poll

December 15, 2005 Meeting

August 10, 2005 Meeting

Modeling the process of photosynthesis is well in hand; there is some evidence that integrated values over 16-day with NASA's MODIS enhanced vegetation index closely match those derived with eddy-covariance measurements.

A recent review by Eric Hobbie (Ecology 2006) suggest that the fraction of photosynthate allocated below ground to symbiotic fungi varies significantly from 0, on fertile soils, to 20% on extremely infertile soils. These observations, combined with experiments that girdle trees in different seasons and on soils of differing fertility (and water status) open the door to much improved modeling of seasonal (2 wk to monthly resolution) variation in allocation above and below-ground. Since shifts in allocation explain large differences in species production under the same climate, additional knowledge will improve all growth models and help explain differences observed among species. Some non-destructive experiments using cold water to halt phloem transport are underway by Kurt Johnsen and Chris Maiser, USFS Expt. Station, Durham, N.C.

Following Dave Hamlin's idea that the Bakuzis Matrix should/could be implemented in R, I spent some time and put one together.  An R package can be downloaded HERE.  Download it to your machine and use the "Install packages from local zip files" option.  The current version is 1.1.

The package includes a small data set including three DFSIM runs on site 100, 120, 140.  Use the example in the help file to see the results:

> library( bakuzis )
> data( dfsim, package="bakuzis" )
> bakuzis( group=dfsim$site, age=dfsim$age, height=dfsim$lht, stems=dfsim$tpa, ba=dfsim$ba, volume=dfsim$volume )
> ?bakuzis

Any suggestions for improvements are welcome.

Here's an example of the output from ORGANON contributed by Dave Hamlin.

After Dave Marshall's talk yesterday I thought I'd do a couple of quick FVS (RV:02.25.05) runs looking at the difference in curve shapes between 5 and 10 year steps with no other modifications. The results of running 2 stands (yeah, VERY small sample...) from Pack Forest:

Curves for TPA, QMD, BA, Top Ht, and SDI are below. Also there are svs files for each stand so you can see what they look like.



What I found was that the curves were the same for 5 and 10 year steps for BR_TRI and seem to be "reasonable".  Differences in curve shape do show up between 5 and 10 year steps in BR_1300_RD leading me to wonder about the effect of running young stands with FVS.

It will be interesting to look at the age distribution of the plots in the database that was used to fit the FVS curves. Maybe with young stands we are pushing the range of the data used for the models and getting strange behaviors. Any thoughts on this?

Greg Johnson made a good beer bet that the maximum SDI produced by ORGANON would vary by site index.  Dave Hamlin sent this PDF file.  He ran site indicies 85, 105, 125, and 145 through the NWO, SMC, and new SMC versions of ORGANON to look at the SDI achieved across site index.  What he found (in the PDF) is that maximum SDI increases as site index increases unless the stand hits the maximum SDI set by the user.  Greg wins the bet, but hasn't found anyone to pay up!

Once a long time ago I had some computer science types working for me and I had them develop a program that would plot each stand property in a "standard" yield table against every other property. All the bivariate properties that had been studied and found to be regularly repeated patterns happened to fit below the diagonal, so the matrix was more compact. The idea for doing all this must be credited to the late Prof. Egolfs V. Bakuzis, U. MN, School of Forestry. So, I named it the 'modified Bakuzis matrix' of stand properties.

I've tested a lot of growth models using it, and not a single one has "passed"  -- i.e., not violated at least one of the interior cell relations that we call 'laws of nature' governing stand development.

I've included an example test of the Prognosis output that was published in a publication form Intermountain Station way back in the 1990s. Can't recall which publication it was as I type this, but I could look it up. You will note that this version of Prognosis for Douglas-fir violates the Sukatchev effect (a time-density relation [Gause 1934, Harper 1977] says that self-thinning will begin to occur earlier on a good site than on a poor site) -- for a few decades. This output is generated by a Mathematica notebook, written by Kevin Nimerfro -- downside is that Mathematica costs about $1000. The input files are simply numbers from tables in the Int. publication. If a cell is blank, the relation is not known to be repeated with any regularity among species

Recently I had a visitor from Denmark, Dr. Mads Jeppe Tarp-Johansen. We shared an office when I worked there in 1997. MJ is an expert Java programmer. So, for room and board for 2 weeks, he wrote a Java applet that does the modified Bakuzis matrix in a very elegant way -- for free -- on your own computer. There is even space for a brief explanation of the regularly repeating forms and a simple equation where appropriate.  See this link for the Java applet.

Gause, G.F. 1934. The Struggle for Existence. New York: Hafner. 163 p.
Harper, J. 1977. Plant Population Biology. New York, NY: Academic Press. 892 p.