i-Tree Suite version 5.0.9 updates available

Issues related to i-Tree v5.x can be posted in this section. 5.x updates and known issues will be posted here also.
Post Reply
azelaya
i-Tree Team
Posts: 372
Joined: Fri Jun 27, 2008 10:50 am

i-Tree Suite version 5.0.9 updates available

Post by azelaya » Wed Sep 11, 2013 1:48 pm

i-Tree Suite v5.0.9 updates are now available for download. i-Tree Streets and the Eco have been updated to address several known issues mostly related to reporting. The update patch can be accessed through the Help - Check for Updates option in most applications.

Details of the update patch can be viewed in the i-Tree Suite Change Log http://www.itreetools.org/resources/con ... ge_log.pdf

Al
A member of the i-Tree Team
Jerry
Posts: 100
Joined: Wed Nov 28, 2007 8:11 pm

Post by Jerry » Wed Oct 09, 2013 1:45 pm

Well this is weird! I updated today from 5.0.6 to 5.0.9, though it took me three times to get it to finish correctly. I opened Streets and confirmed that the running version was 5.0.9. Then I opened an existing sample inventory project that has run fine in every version since 2007, and it seemed to be OK. When I checked the report on population, however, the total estimated population was given as 2,996. Problem is, every other version has reported 3,811!

I only installed Streets and Eco, if that makes any difference. I also noticed a message about updating project files somewhere along the way. Has anything like this been reported before on this or earlier versions? Suggestions?

I am going to try bringing in the original project file and see if that makes any difference.

TIA
Jerry
Posts: 100
Joined: Wed Nov 28, 2007 8:11 pm

Post by Jerry » Wed Oct 09, 2013 2:07 pm

Didn't make any difference. Guess my next move is to create a new project from the original data, huh?
azelaya
i-Tree Team
Posts: 372
Joined: Fri Jun 27, 2008 10:50 am

Post by azelaya » Fri Oct 11, 2013 8:50 am

This may be related to an error discovered by user in Thailand who discovered a discrepancy with the population summary table. The population summary report was not accounting for sample projects with zone stratification correctly. I believe that individual zone population estimates were correct but the summary total was incorrectly combining all segments as one sample generating a different population total incorrectly. i-Tree 5.0.8 & Streets 5.1.0 integrated a major rewrite of most Streets reports and addressed the incorrect summary report. Therefore, updating your version and running the project may affect the old estimates.
A member of the i-Tree Team
Jerry
Posts: 100
Joined: Wed Nov 28, 2007 8:11 pm

Post by Jerry » Mon Oct 14, 2013 6:56 am

You mean that all the work I have done under contract to municipal clients over the last 7 years has been off by 25-35%?!

I am going to check the Tiger-Line mileage that we sampled and do a simple proportion:

sampled trees/sampled miles = total trees/total miles

That should give me the ballpark figure I need ATM.
Jerry
Posts: 100
Joined: Wed Nov 28, 2007 8:11 pm

Post by Jerry » Mon Oct 14, 2013 10:16 am

To follow up on this: we originally sampled 6% of the blocksides, and got the following results:

total mileage: 3.34
number of trees: 284

Since the total street line mileage using T-L files is 51.4, running the simple proportion yields 4,370 trees as an estimate for the entire city.

That may be somewhat high, but the estimate of earlier versions is a lot closer to it than what the newer versions are producing.

Thoughts?
User avatar
dellings
i-Tree Team
Posts: 31
Joined: Thu Aug 21, 2008 2:12 pm

Post by dellings » Mon Oct 14, 2013 12:02 pm

Jerry, the estimations in Streets are not based upon the number of linear miles. The basic calculation is as follows:

Est # of Trees in Zone = # of Trees Sampled in Zone / # of Segments Sampled in Zone * Total # of Segments in Zone

In versions prior to Streets 5.1.0 the total number of trees in the population was calculated using:

Total # of Trees in Pop = # of Trees Sampled in All Zones / # of Segments Sampled in All Zones * Total # of Segments in All Zones

The issue with calculating the total number of trees via this method is that it ignores the stratification of the zones. The resulting estimate will not match the total number of trees estimated for each zone. This issue is prominent throughout all the reports in i-Tree Streets prior to 5.1.0
Jerry
Posts: 100
Joined: Wed Nov 28, 2007 8:11 pm

Post by Jerry » Mon Oct 14, 2013 12:26 pm

Thanks. I knew that Streets did not calculate this way, but was just trying to get another perspective on the population question.

At the population level, I believe that using 6% of the mileage in the little proportion I cited above should produce a valid estimate as well, if the initial distribution of samples was random and if the trees are at least reasonably evenly distributed. The latter is highly unlikely in many urban sites, but a high enough sample rate should take care of all but extremely skewed distributions. The spatial distribution of my 6% sample in this case was fairly good, as the attached shows.

So is your bottom line that the new figure is fine and all the older ones were faulty? Has anyone tested this taking random samples from complete inventories to compare the sample estimate with the true value? This is of more than academic interest to me, since I have a contract proposal to submit in the near future.

Thanks.
Attachments
SamplesSpace.png
Distribution of sample segments
SamplesSpace.png (27.69 KiB) Viewed 18864 times
Jerry
Posts: 100
Joined: Wed Nov 28, 2007 8:11 pm

Post by Jerry » Mon Oct 14, 2013 2:49 pm

OK, hold it. Those spreadsheet numbers included planting sites, and if I take them out the numbers become as follows:

total sample mileage: 3.34
number of trees in sample: 172

That gives me this population estimate of trees only for the whole city:

(51.4 x 172)/3.34 = 2,647

So now the new STREETS estimate is the closer one, and I will use it. Since the proposal will be for a complete inventory, it will be interesting to see how things shake out.
User avatar
dellings
i-Tree Team
Posts: 31
Joined: Thu Aug 21, 2008 2:12 pm

Post by dellings » Mon Oct 21, 2013 2:24 pm

Jerry,

Technically speaking it's not that the old values were wrong, its that they were inconsistent with the values shown on other reports. The purpose of stratification, or zones, is to increase the accuracy of population estimates by dividing the population into additional sub-categories. One reason you might want to stratify your data is to reduce the impact of tree density between two or more strata. For example, residential zones are most likely to have more trees per street segment than industrialized zones. Additionally, some trees may not thrive well in an a more industrialized environment. Stratification allows you to reduce the impact of these effects.

If you do not wish to stratify your study area, then you can use a single zone that covers your entire study area. If you remove the stratification from your one of your old projects, the results seen on the reports would match the old values. *Note: street segments must be unique prior to combining the zones.

The changes made to these reports will not have any effect on a full-inventory project since no form of estimation is being done. A comparison between a full-inventory project and a sample derived from it should produce results that are more consistent than before due to the points stated above.
Post Reply