More on error analysis and related topics
   Nov 27, 2004 06:16 PST 
     For our new members, we should explain what all the fuss is about over tree measuring and associated errors. So at the risk of redundancy in the extreme, for our new members, what follows is some ENTS history.

    About 15 years ago, I set out to document the old growth forests that I was exploring in the Berkshires and Taconics of western Massachusetts. I wanted to record the dimensions of significant trees I was encountering as part of the documentation. I borrowed a Haga altimeter (a brand of clinometer) from Harvard Forest and proceeded to measure lots of trees and convey the results to organizations and people who were interested in old growth. Along the way I met a timber framer-architect-surveyor named Jack Sobon. My tree heights were often higher than his and the numbers for one particular tree, a sugar maple at the base of Todd Mtn, that I had measured sounded impossible to him. So we measured the tree with his transit and my clinometer-based measurement proved to be 21 feet off. That opened my eyes, especially considering that on different occasions, a dozen or so other folks had measured the tree with me and those folks were not amateurs. They were foresters, plant ecologists, forest ecologists, naturalists, etc.

    I sat down and began analyzing the possible sources of error that led to the sizable error. With mathematics as my minor in college, I wasn't timid about tackling the problem. I began by considering the actual shapes of trees and where their high points might be. I examined lots of trees and drew lots of diagrams and quickly came to realize that the fatal assumption that I had made as well as the others was that the spot in the crown I was measuring as the top, that from a distance looked like it was over the base, was in fact often 5 to 10 feet, and occasionally much more, horizontally out from the base. But traditional calculations treated as though you were always measuring a point over the base. That realization and discussions with Will Blozan spawned our methods of crown cross-triangulation. The adjustment helped us greatly. The method is described in a book Will, Jack, and I wrote entitled: "Stalking the Forest Monarchs - A Guide to Measuring Champion Trees". We covered use of the laser and included advanced methods for measuring average crown spread. We extended our reach through e-mail and the internet to include numerous diagrams and explanations that both Will Blozan and I would convert American Forests and the coordinators of the state-level champion tree programs. Will Blozan and I knew long ago that we had an unusual situation to deal with. For lack of a laser device that could measure distance to an actual point in the crown, methods had been adopted that worked only for idealized trees- the vertical telephone pole in the level parking lot analogy and the proponents of the tangent-based methods were not inclined to give ground. They just couldn't accept that they were using flawed techniques. American Forests was a particularly tough nut to crack. I've since given up on them.

    We're making our greatest inroads with the state coordinators, which Georgia coordinator Will Fell suggested about 2 years ago. Perhaps the biggest boost came for us was when Dr. Lee Frelich joined ENTS and began helping us promot and refine the ENTS methods. Others in academic circles who had previously balked at coming on board with us listened to Lee. Dr. Tom Diggins has added more academic horsepower.

     Since Lee and Tom came on board, a second big boost has come as we began picking up mathematicians like John Eichholz and Howard Stoner. Ed Frank must now be counted in among the heavy weights. With experienced foresters like Don Bertolette, Russ Richardson, Will Fell, and others on board, the circle is being closed.

    Are we making a big to do over nothing? Only if accuracy is unimportant and that would be a hard sell for anybody who is scientifically inclined. To suggest that we should be sure to measure circumference (and compute diamter) to the nearest tenth of an inch, at a prescribed distance above the base, but now worry if we're off by tens of feet in another measurement would be bizarre to say the least. And we do not exaggerate when we point out measurements that have been in error by tens of feet.

    On many past occasions, either Colby Rucker, Will Blozan, myself, Dale Luthringer, or others pointed to tree height errors made by measurers using the % slope method that were between 20 and a whopping 60 or more feet. There is no way that errors of such magnitude can be acceptable to a serious endeavor and there in lies the answer. Tree measuring for the champion lists wasn't really a serious endeavor for those in professions that routinely dealt with tree measurements for economic purposes. And there was no compelling reason why they should have felt differently. However, the same is not true for ENTS.

    One of our primary purposes is to profile the growth capabilities and attendant limits of eastern tree species across their natural ranges. That requires we take accuracy in our measurements seriously. Measurements for big tree contests are peripheral to our interests, though we do not shy away from the fun of hunting record trees. But we have no compelling interest in pushing the champion tree point system. Individual measurements of circumference, height, and spread, trunk and limb volume, percent of crown, crown area, point of major branching, rate of taper, etc. all these are of interest to us.