Study Attempts to Predict Max Tree Height Across U.S.

Reports and articles related to remote sensing, LIDAR, GIS, and other mapping applications.

Moderators: edfrank, dbhguru

Post Reply
User avatar
pauljost
Posts: 30
Joined: Sun Mar 14, 2010 1:24 pm

Study Attempts to Predict Max Tree Height Across U.S.

Post by pauljost » Tue Jul 19, 2011 9:15 am

I don't know if this has been posted to the ENTS BBS yet, if it has, I missed it and apologize for the duplication.

MIT, U. of Maryland, and the Santa Fe Institute recently published the following study that attempts to predict maximum tree heights across the country based upon USFS FIA data, weather data, and other parameters. Of course, they omitted known maximum tree heights from other sources, such as ENTS, that actually seek out maximum height data. Read for yourselves:

http://web.mit.edu/newsoffice/2011/tall ... -0718.html
http://www.plosone.org/article/info%3Ad ... ne.0020551
journal.pone.0020551.pdf
(723.61 KiB) Downloaded 106 times
The observations and predictions both have great errors from known maximum height measurements. The study, based upon a terribly limited data set, has questionable use. It would have been interesting to see the results if actual maximum height data had been used.

Paul J.

User avatar
Will Blozan
Posts: 1153
Joined: Fri Mar 12, 2010 7:13 pm

Re: Study Attempts to Predict Max Tree Height Across U.S.

Post by Will Blozan » Tue Jul 19, 2011 5:35 pm

Paul,

It baffles me how in this day of internet resources that no one contacted ENTS or the WNTS counterparts regarding our work with north American trees, or more specifically our recent work with LiDAR groundtruthing. It is obvious the model is way off and that fact they used no reliable resources to qualify their predictions places this paper directly in the recycle bin.

Will

User avatar
Steve Galehouse
Posts: 700
Joined: Mon Mar 08, 2010 11:50 pm

Re: Study Attempts to Predict Max Tree Height Across U.S.

Post by Steve Galehouse » Tue Jul 19, 2011 9:58 pm

Will-

I agree entirely.

Steve
every plant is native somewhere

User avatar
AndrewJoslin
Posts: 408
Joined: Sun Mar 14, 2010 7:36 pm

Re: Study Attempts to Predict Max Tree Height Across U.S.

Post by AndrewJoslin » Thu Jul 21, 2011 7:43 am

I think the headline in the MIT News article oversold what the study has tried to accomplish to-date. My impression is they're trying to establish a foundation methodology. For example they're using composite tree characteristics, essentially a generic tree. This seems to be a model for analysis not something that would actually provide accurate results without applying characteristics and accurate potential height numbers for individual tree species. The writer mentioned that in the future they could apply the analysis for specific tree species.

This paragraph from the paper pretty much says they're not expecting accuracy until they get better source data for individual tree species (the bold/color emphasis is mine):
We employ a single generalized tree across a range of environments without specific knowledge of local plant functional types commonly used in previous models [7]–[9], [34], [35]. In doing so we sacrifice accuracy at the local scale but gain a simple understanding of the average variations across environments. In the context of resources our framework lends insight into the mechanisms underlying deviations from the allometric scaling laws where, for example, we are able to show how different tree traits are suited to a given environment and predict the temperature-based variation of stomatal density (Fig. 5). These variations in turn modify the size-based scalings for an individual tree species (see Supplement S1). Ongoing work is beginning to understand departures from the basic zeroth order allometric scaling laws [24], [27], [36], [37], however, it should be noted that the zeroth order theory has yet to be coupled with environments in order to test its predictive power. Our work provides another means for expanding the basic allometric scaling laws to encompass features that are relevant to more detailed modeling efforts.
Might be worth it for the ENTS to contact the author and point them to accurate height and volume data for specific tree species.
-AJ

User avatar
Lee Frelich
Posts: 155
Joined: Mon Mar 15, 2010 3:24 pm

Re: Study Attempts to Predict Max Tree Height Across U.S.

Post by Lee Frelich » Thu Jul 21, 2011 8:28 am

Andrew:

A small group of us has discussed this paper among ourselves-but now that it has gotten on the BBS, here are my comments. Although there are some good ideas, this is obviously a horrendously flawed study.

1. The FIA (Forest inventory and analysis data from the Forest Service) does not yield maximum tree heights, instead its just a random sample of the heights of trees across the landscape, with a variety of stand ages. Also, and the locations of the plots the FS gives are fuzzed within one mile of the true location, so one cannot be sure that the locations are within 4 km and 100 m elevation of a weather station as they state in the paper.

2. They don't define maximum tree height in the observed data. Obviously one can get a number of different maximum heights from field data depending on the definition used.

3. They failed to notice that tree heights get shorter in warmer climates that are dry in summer (west coast), but the opposite pattern occurs in summer wet climates (eastern US). Therefore, their predictions for pattern of tree height with latitude is backwards in the eastern US. Its so obvious that this is the case in their figure 1, even with the flawed field data they used, that one wonders what they were thinking when they claimed that their predictions had some short of validity (and the bizarre method they used to compare predictions with field data allows such huge departures of the predictions from field data that it is useless). This also means that their predictions for changes in tree height with climate change in the east, which has most of the forest area in the U.S., are backwards compared to what is actually likely to happen (one of those nasty sign errors that also plague published papers that deny global warming).

4. They state that they underpredict tree height in arid environments, apparently failing to notice that trees in those regions grow in riparian environments, so are not as constrained by amount of rainfall as their model predicts.

5. All sorts of wild claims are made without support. For example, that one can infer the size structure of a forest and biomass by knowing the height of the tallest tree, in addition to reconstructing the structure of paleo vegetation by knowing trunk diameter of a fossil tree. Apparently they don't know that height:diameters ratios also change across the landscape.

Obviously the authors just do math and don't relate well to the real world. I have been thinking about whether to respond to the paper with a critique, ignore the paper, or try to do something better.

Lee

User avatar
AndrewJoslin
Posts: 408
Joined: Sun Mar 14, 2010 7:36 pm

Re: Study Attempts to Predict Max Tree Height Across U.S.

Post by AndrewJoslin » Thu Jul 21, 2011 9:23 am

Thanks for the clarifications Lee. I didn't doubt that Will and others have very good reasons for being skeptical but tried to take the paper at face value and understand it as best as a I can not being a scientist.

I'd say do it better yourself :-)
-AJ

User avatar
AndrewJoslin
Posts: 408
Joined: Sun Mar 14, 2010 7:36 pm

Re: Study Attempts to Predict Max Tree Height Across U.S.

Post by AndrewJoslin » Thu Jul 21, 2011 9:26 am

Lee Frelich wrote:Obviously the authors just do math and don't relate well to the real world. I have been thinking about whether to respond to the paper with a critique, ignore the paper, or try to do something better.
Seriously though it might be worth providing constructive feedback, nothing to lose. At least it's one way to get the authors aware of the big world of excellent data and resources beyond what they seem to be aware of.
-AJ

User avatar
Neil
Posts: 143
Joined: Sat Mar 13, 2010 2:17 pm

Re: Study Attempts to Predict Max Tree Height Across U.S.

Post by Neil » Fri Jul 22, 2011 7:01 am

thanks Lee for seriously breaking this down.

related to the discussion on peer-review: there is a way papers are ranked/voted on as popular/worthy, by citing it. of course, weak papers can get many citations because they are so bad that folks spend a lot of time showing why it is wrong. in this case, it might be best to let this one die a slow death. if, however, people pick it up and use it without critical thought, then it is likely best to use precious time pointing out the flaws.

neil

Post Reply

Return to “Remote Sensing, LIDAR, and GIS”