Sent: Wednesday, September 28, 2005 10:30 AM

Subject: More testing of RD 1000

Hi All:

The attached spreadsheet provides a detailed look of the challenge one faces in using the RD 1000 to model trunks. I took two sets of readings, one at a level distance of 48 feet and one at 64.5 feet. Note the different pattern of values obtained for roughly the same spots on the trunk of the silver maple being modeled at a 48-foot level distance (blue lines) from the target as opposed to the 64.5-foot distance. Distances to the target closer than 55 to 60 feet definitely produce a pattern of over-measurement of the diameter with the RD 1000. That conclusion is no longer in doubt. Unfortunately, the trunk is most visible at the closer ranges, and regrettably, use of the magnifier at greater distances as a compensation does not work well. The magnifier definitely produces errors on the high side. The big question is can we develop a calibration table to compensate for the pattern of errors that I am documenting.

By contrast to overages occurring at close range, level distances from the trunk of over 95 feet away can lead to under-measurement of the diameter. The magnitude of this effect is harder to get a handle on, though, because being farther away from the tree makes read the scale and adjusting it to get the best fit to the trunk much harder. In addition, the value of an expansion or contraction click invokes a wider interval, i.e. being off one click carries a greater penalty at greater distances.

From tests conducted thus far, the ideal level distance away from the trunk appears to be from 64 to about 82 feet to get accuracy at eye level within the range advertised. The big question is whether or not that at the 64-82-foot distance range, scans far up the trunk produce increasingly larger errors. Testing this possibility is where someone needs to be up in the tree taking periodic circumference measurements from aloft.

When Will climbs the Joseph Brant Pine in MTSF on Oct 15th, maybe we can mark several spots at different heights on the trunk. If we can put orange flagging on the trunk at each point of measurement, I'll have some realistic calibration targets. The tree is well off any trail in steep terrain and obscured from any distant vista, so any flagging we might use won't be seen. I and others can go back to the tree on different occasions and re-shoot the flagged points from different level distances and examine the patterns of values. Such a test will provide the most realistic field test of the instrument, a test that combines distance, angle of inclination, visibility, and trunk shape irregularities.

My current belief is that I've erred slightly on the conservative for the last 7 or 8 trees that I've modeled and reported to you all. However, I'm shaky on that conclusion. The testing of the RD 1000 will need to go on for several more months before I am willing to trust my results. I'm hoping that my mathematician compadre John Eichholz will have time in the off season to help me with the analysis of sources of error and with the development of calibration tables. Without them, the errors committed would be too great. However, I'll have some handy rules of thumb to use before that such as further affirmation of the 64-82 foot rule.

As a final observation, nothing less than exhaustive testing of the RD 1000 is required to make this instrument usable for serious trunk modeling. This I can state categorically for at least my instrument. One cannot roam the woods pointing and shooting and hope to get accurate results for the volumes of individual trees.

Bob

Database Developer and Systems Analyst

Information Technologies

Attachement:  Janis Silver Maple RD 1000