Foliage Calculation for Summary Statistics

In order to compare foliage for different foliage measurements on the DNR Leave Tree summary statistics, we needed to convert Foliar Units to Percent Cover.  This page explains how we did that conversion.
 
The 1000 Year Chronosequence Study records foliage data in two different ways. Percent Cover is a relative measure that describes the proportion of each branch from which foliage grows. Foliar Units are an absolute measure of quantity of foliage. The graph below shows percent cover along the vertical axis and branch length along the horizontal axis; they are uncorrelated.
Percent Cover vs Branch Length graph: This graph shows foliage percent cover (vertical axis) and Branch length (horizontal axis) for those trees that record it in the Trout Creek, Carbon River, and Cedar flats on the Thousand Year Chronosequence Study.

In order to display all trees in the catalog in easily comparable ways, we needed to convert foliar units to percent foliage. A very imprecise conversion was used.
Foliar Units vs Branch Length Graph: This graph shows foliar units (vertical axis) and Branch length (horizontal axis) for those trees that record it in the Trout Creek, Carbon River, and Cedar flats on the Thousand Year Chronosequence Study.
Foliar units are weakly correlated with branch length (see graph above), so a relative value can be calculated based on foliar units divided by branch length. This relative value is scaled to percent cover using a site-specific constant.
Percent Cover = (Foliar units / (Site constant * Branch length)) * 100
The site constant is determined by calculating the mean percent cover at all branches on all douglas fir trees at the site and comparing to the mean (foliar units / branch length). Because foliar unit data are only present for douglas fir trees, we do not need to worry about other species.
Site constant = (Total foliar units * 100) / (Length of all branches with foliar units * Mean Percent Cover)

Site Constant
Trout Creek 3.7
Carbon River 4.0
Cedar Flats 3.8

While we feel the converted values are good enough for a low-resolution visualization, we do not think these data are suifficiently accurate for analysis or use.