Hi there
I am trying to replicate the iTree Eco tree count statistics. I have the plot data and areas of land use type for London. But with simple arithmetic I can not recreate the numbers produced with iTree Eco software. I have also tried to correct using percent of plot measured. Please can someone point me at a reference for calculating these values or describe the method.
Regards
Reference for method to estimate tree count
Moderators: mbinkley, iTree Team

 iTree Team
 Posts: 166
 Joined: Fri Dec 21, 2012 7:45 am
Re: Reference for method to estimate tree count
Thanks for the question.
The basic method that iTree Eco uses is stratified random sampling. I am not sure what data you have so it is difficult to describe the calculations you would need to perform. Also, I am not sure of the exact details of the London sample design as it is a little atypical with the inner/outer London summaries.
That being said, it is typically just an area weighted average where the areas are for the strata used in the project. The basic steps would be something like:
1. Calculate the total number trees measured within each stratum. i.e. count the number of measured trees on all plots within each stratum.
2. Calculate the trees per hectare in each stratum by dividing each value from step 1 by the total area of all plots measured in that stratum. Be sure to use all plots including those that don't have trees and account for plots that were only partially measured.
3. Calculate the trees per hectare for the whole project by taking the area weighted average of the stratum values from step 2.
4. Calculate the total number of trees by multiplying the value from 3 by the total study area hectares (i.e. the sum of the area of each stratum).
Please note, you need to have the plot data, the stratum each plot was assigned to, and the area of each stratum to replicate that calculations used in the London study.
If you are looking for a more rigorous description of stratified random sampling and methods for estimating sample means and standard errors there are many statistical references including, https://en.wikipedia.org/wiki/Stratified_sampling
If you need additional information please email us at info@itreetools.org. We have a large document summarizing all iTree methods that is currently under review. It has a section that covers the statistical sampling methods that could be shared in draft form.
Thanks,
Jason
The basic method that iTree Eco uses is stratified random sampling. I am not sure what data you have so it is difficult to describe the calculations you would need to perform. Also, I am not sure of the exact details of the London sample design as it is a little atypical with the inner/outer London summaries.
That being said, it is typically just an area weighted average where the areas are for the strata used in the project. The basic steps would be something like:
1. Calculate the total number trees measured within each stratum. i.e. count the number of measured trees on all plots within each stratum.
2. Calculate the trees per hectare in each stratum by dividing each value from step 1 by the total area of all plots measured in that stratum. Be sure to use all plots including those that don't have trees and account for plots that were only partially measured.
3. Calculate the trees per hectare for the whole project by taking the area weighted average of the stratum values from step 2.
4. Calculate the total number of trees by multiplying the value from 3 by the total study area hectares (i.e. the sum of the area of each stratum).
Please note, you need to have the plot data, the stratum each plot was assigned to, and the area of each stratum to replicate that calculations used in the London study.
If you are looking for a more rigorous description of stratified random sampling and methods for estimating sample means and standard errors there are many statistical references including, https://en.wikipedia.org/wiki/Stratified_sampling
If you need additional information please email us at info@itreetools.org. We have a large document summarizing all iTree methods that is currently under review. It has a section that covers the statistical sampling methods that could be shared in draft form.
Thanks,
Jason
A member of the iTree Team
Re: Reference for method to estimate tree count
Thanks Jason
Are there any filters applied e.g. are dead trees counted (and how are they labelled), multistem trees counted as 1 or x number of stems etc. I am getting a ballpark correct figure (~89M) but would be much happier if I could get exactly the same as reported by the software.
My aim is to then filter trees by size and estimate number of e.g. larger trees.
Thanks again
Are there any filters applied e.g. are dead trees counted (and how are they labelled), multistem trees counted as 1 or x number of stems etc. I am getting a ballpark correct figure (~89M) but would be much happier if I could get exactly the same as reported by the software.
My aim is to then filter trees by size and estimate number of e.g. larger trees.
Thanks again

 iTree Team
 Posts: 166
 Joined: Fri Dec 21, 2012 7:45 am
Re: Reference for method to estimate tree count
Dead trees would be included and multistem trees are counted as a single tree. The reasoning with including standing dead trees is that they are still storing carbon and therefore still providing some benefit.
As far as replicating the software exactly you will want to make sure you are using the exact same plot and strata areas since those get applied as multipliers in all your calculations. There is also likely some rounding error in both the software and external calculations. I would say if you get within a percentage point or two that is probably what you are looking at.
To your filtering idea you could also do that in the software. You could create a copy of your project, open the "Tree" data table and sort by clicking on the appropriate column heading, then select the rows you are not interested in an delete them. Holding ctrl or shift while you click works the same as in Excel for selecting single rows or multiple rows respectively. I've done this a couple times to create summaries of just a single species of interest.
Thanks,
Jason
As far as replicating the software exactly you will want to make sure you are using the exact same plot and strata areas since those get applied as multipliers in all your calculations. There is also likely some rounding error in both the software and external calculations. I would say if you get within a percentage point or two that is probably what you are looking at.
To your filtering idea you could also do that in the software. You could create a copy of your project, open the "Tree" data table and sort by clicking on the appropriate column heading, then select the rows you are not interested in an delete them. Holding ctrl or shift while you click works the same as in Excel for selecting single rows or multiple rows respectively. I've done this a couple times to create summaries of just a single species of interest.
Thanks,
Jason
A member of the iTree Team