by Kim Iles

I have been working with Variable Plot Sampling, 3P, and general forest inventory for more than 40 years now.  In that entire time, I cannot remember a cruise compiler programmer ever asking me what he should plan for when designing the next program.  The view was always through the rear-view mirror, and usually backward by at least 30 years.  While it is true that waiting until something has a track record is a good idea, perhaps we could look forward at least a little bit.  It would be great if we could use our data to analyze what would work well.  If you have the chance to influence these folks, here are a few ideas they might think about – for Fixed or Variable Plots.

Keep better precision.  I know that your BAFs are normally to an even number, but what happens when you want to use that prism for a Metric computation?  What if every site index changes by just 3.2%?  Rounding numbers or minimizing digits might have been needed when the computer was 1/100,000th of the current speed and only had 16K memory – but those reasons died long ago. 

Use a BAF (or plot size) for every tree.  In most cases, you only need one for each inventory, but there are exceptions.  One easy way to adjust a current set of data by 1.53% is to change the Fixed Plot size or BAF by that amount - and recompile.  You get trees of the same description, but more or fewer of them - so the total is corrected by the amount you wanted.  This is one of the best ways to use old data to describe a land area, while adjusting the overall total to a new answer.  If you want to adjust the Cedar up 5.1% and the Hemlock down 1.4%, you adjust plot size or BAF by species.  Do some trees have a different volume for some reason? – adjust those trees individually.  If every tree has a BAF or plot size, these techniques are easy to apply.  This should be available to users.

Pump out the partial tree information so the client can work with it on the side.  Dropping it into a flat file or an EXCEL™ format would perhaps be the most general way to do this.  In many cases, the question of “what is the difference in ___?” comes up.  If you had individual tree characteristics such as diameter, height, crown, % decay, value and volume - with indications of the location (which brings in data from your GIS source) many of these questions could be examined. 

Get ready for metric.  I know, it will never happen in your lifetime - but stranger things have come to pass.  You don’t need the computations yet, but set up the data entry to handle that, so that is not disrupted any more than is necessary when the time comes.

Anticipate subsampling for trees.  In general, we far oversample trees for stand volume.  To begin with, make sure that BigBAF sampling or any equivalent is allowed (since it has been nearly 40 years since it was introduced, it’s time to catch up).  Keep the Basal Area and VBAR statistics separate and well reported.  Some graphics would be nice, too.  Do the user a favor and suggest the right ratio of counts vs. measured trees indicated by that compiled cruise.

In other cases, subsampled trees can be weighted by a multiple - just as we correct for edge effect by doubling the count for some trees.  If a tree represents 3 other fir, it can just be treated (duplicated) as 3 trees instead of one.  A simple multiplier for each tree will do it.  While it would be nice to be able to handle a situation where a tree represents 2.56 other trees that is harder; but it would solve most problems if a simple multiple was used.  It needs very little computer code to allow this weighting, and a simple data entry change to get ready for it.

More flexible groupings of trees.  You might know the sampling error for your Spruce, and perhaps for the Pine – but what’s the sampling error for the intermingled combination?  If you could call them both a single species, you could find out (and no – there is no correct way to combine individual sampling errors for intermingled groups).  Same comment for splitting a species.  One of the easiest ways to get results for a special category is to call it a new species
(so allow easy compilation of some “invented” species codes for any new cruise).

Allow Non-compilation trees (or plots).  There are many reasons to gather data on trees or plots that do not meet current inventory standards.  Don’t skip all those trees, or make an effort
to throw away that data later, just tick a category that computes the tree data but does not add it into the plot volumes or other summaries for that inventory.  Later, if you want to see the effect, just change the category.  When the data is dumped, those compiled tree volumes or values might be very useful to examine. 

This also works for “close” trees that are just out.  When you visit a Variable Plot PSP next time, it will help predict which ones will be “in” as they grow in diameter. 

Which brings us to distance.  Be able to note the distance to the tree.  A few forms of Variable Plot Sampling can use that distance in compiling better plot volumes.  There are other analytic reasons to keep track of distances to trees.  Get it into the data entry. 

Most people get the Lat/Long coordinates of the plot.  Make sure it is in the data so you can recover other information about the plot from GIS sources, and for plotting results. 

Unassigned data fields should be available in the data entry for temporary and future use (make sure a field title can be entered and stays with the data).  The compilation might be modified later if necessary, but the entered data itself is often enough for special studies. 

A great quality control function would be provided by hand-held systems if they could compile data on the spot, so the cruiser would know what he was looking at in terms of truckloads, volume, value, or other summary data.  This would train cruisers to make better estimates, and allow much more efficient sampling systems to be used.  3P sampling is the most obvious use of a cruisers ability to estimate tree or stand characteristics.  There are many other sampling systems that could use this ability.

If you could compile on the spot you could see the effect of changing one or more diameters, defect calls, or height measurements.  This is exactly what check cruisers should be able to do as they check a plot for either training or adjustment reasons.  It is a shame that cruising standards are still set by guessing, rather than determining what really matters to the compiled answers or to the use for the data. 

Many of these approaches have been used (and some of them 30 years ago), but will not be common for years.  Because they have not been anticipated in computer compilers holds us back from useful change.  In some regions of the country, they are barely out of fixed plots.

Why do we need such things?  At present you hear a lot of “well, that probably wouldn’t work anyway– and since we can’t check, I am really sure it wouldn’t work”.  If you have the ear of a compiler programmer some time, you might suggest some of these things.  We have been dragging computer programs behind us for a long time.  It would not be that hard for them to get ahead of the game for a change.  Let’s get out of the rear-view mirror.

Originally published February 2018

Return to Home
Back to
Regular Article Index