Handling bias in the sine measurements

[teixeirak]

@mcgregorian1, first, consider if you really want to include the sine method data. Assuming you'll include only one value per tree, do you gain much by including it? If including, we should calculate and apply some correction factor (as mentioned in issue #25).

The best approach is probably this: for each tree... 1- estimate how much each tree you remeasured may have grown since the first measurement based on its change in diameter (requires using an allometry, so there's a bit of circular reasoning. I'd apply the all-species allometry excluding sine data). ∆h would be height estimated based on current dbh - height estimated based on dbh at time of first height measurement 2- add ∆h to the original measurement, h_0 (h' = h_0+∆h)
3- calculate the % difference between sine height measurement and h'. Then, for all trees... 4- average % bias to get a bias correction factor, 5- apply bias correction factor to to all sine measurements.

[mcgregorian1]

There are 57 measurements that include the sine method. Based on these two graph comparisons, it seems that including them doesn't help out too much, and you're right, it would require bias corrections like you mentioned. I'm fine leaving them out (and thus not doing the corrections), but I'd use them anecdotally as representation for what the tallest heights should be (like I found today in #25).

Do you agree?

Excluding the sine data gives us

Including the sine data gives us:

[mcgregorian1]

After doing a search, I found the following. None of these mentioned actually applying a correction, so I think we're ok to include the measurements as is. Thoughts?

Mention using Nikon ForestryPro, no mention of changing method/applying corrections

• Klingberg et al 2017
• Chen et al 2017
• Kunca and Yosef 2016
• Bapureddy et al 2015 (used for primates, not trees specifically)
• Brahma et al 2017
• Bauters et al 2019
• GMU thesis from 2018
• Moriarty 2018 poster
• Cushman et al 2014 Helene is on this paper for BCI, they used Nikon Forestry 550 with the sine method and cited Helene's 2013 paper, but did not apply corrections or say anything about error associated with the method

Mention using Nikon ForestryPro / Nikon Forestry 550 (same capabilities) and acknowledge errors

• Kearsley et al 2017 cite Helene's paper and say we recognize there's error but it's beyond our control
• Cuni-Sanchez et al 2017 also cite Helene's paper
• Rutishauser et al 2013

Mention using Nikon ForestryPro, used tangent method

• Pohl and Loong 2016 for oil palms
• Harwood et al 2018 for birds

[teixeirak]

Let's go ahead and keep it in.

[mcgregorian1]

I'm going to close this issue since we decided to keep the sine measurements as they are.