Determining Sample Size for Street Tree Sample Survey

[Countin'Trees]

I'm a Planner who is coordinating our efforts to conduct a street tree sample survey. We have about 1600 miles of applicable streets comprised of about 35,000 street segments. We also have Ikonos imagery of the City.

I have looked at iTree and am considering that software for our needs.

Right now I'm struggling with a few questions:

1) What are the benefits of a pre-stratified random sample - what is the best methodology for pre-stratifying?

2) We have a City street centerline and the census tiger street centerline. Then tiger has fewer segments and less mileage. Both contain "streets" such as limited access roads and roads internal to Parks that don't have streets that would have street trees. How do we decide which to use and find a method to weed out streets without street trees in a way that doesn't bias our sample?

3) The primary information we are seeking is the total count of street trees. While all the other data will be important this total count is most critical. We would like to report the final estimate of total street trees with a confidence interval and a confidence level such as, "the study results indicate that there are 250,000 trees (+/-16,000)(at a 95% confidence level) How can we determine the correct number of streets to sample to achieve a given confidence interval and level?

Thanks

[i-Tree Support]

We are not statisticians, but here is some basic orientation on these questions as they apply to components of the i-Tree Software Suite.

1) STRATUM is the street tree assessment tool that allows for sampling based on simple random sampling protocol.. The technical definition of that is a sampling design in which n distinct units are selected from the N units in the population in such a way that every possible combination of n units is equally likely to be the sample selected. Simple random sampling is the only method that can be used for sample inventories in STRATUM.

Your sample can be sub-divided by the management zone of each tree recorded , which allows you to report on the structure, function and value of populations of trees within discrete areas of the community. Typically, management zones are defined in a way that is meaningful to the City: existing political boundaries (e.g., districts or burroughs), neighborhood age, land use or character.

2) If your objective is to evaluate trees on streets that the City maintains, then there will be no bias if you eliminate streets that the City does not maintain. But you can’t eliminate streets without trees that the City does maintain. City-maintained streets without trees must be kept in the sample or you will be unable to obtain an accurate estimate of total tree count by zone or citywide. For sample inventories, STRATUM requires that you enter total number of street segements for each zone, if stratified, or citywide if unstratified.

3) i-Tree components only report standard error associated with the estimates. With additional equations, confidence intervals can be constructed based on the standard error and other parameters such as the sample distribution and your desired level of precision . We suggest you consult a statistician or standard statistical text to guide you in this analysis.

From the i-Tree User's Manual:

"For STRATUM, sample inventories are generally a 3-6% sample of total street segments, depending on community size and variation from segment to segment, and will produce about a 10% standard error for the total number of trees citywide. The following, general guidelines can be used as a starting point for determing sample sizes in communities based on human population sizes:
• For communities with less than 50,000 persons, sample size is 6% of total street miles
• For communities between 50,000 and 150,000 persons, sample size is 5%
• For communities between 150,000 to 250,000 persons, sample size is 4%
• For communities over 250,000 persons, sample size is 3%
It is important to note that all communities differ in their tree density, street miles and population characteristics. Therefore, no single sampling intensity will work uniformily for all communities of a similar size. Ultimately, it is up to the user to determine sampling size and an acceptable level of error based on how the results will be used."

Hope this helps.

[Countin'Trees]

Can you share what the formula is that STRATUM uses to compute and report the "standard error?" We have 1650 miles of streets that are municipally maintained.

Also - Are there any samples of street tree surveys that can be shared?

Thanks-

[smaco]

No Problem, see attached for the equations STRATUM uses along with an example of how they are used.

For an example of real project results based on i-Tree sampling protocols and STRATUM reports, see the Minneapolis Street Tree Resource Analysis report: http://itreetools.org/resource_learning ... s_mbca.pdf.

Note: We know many communities have used i-Tree tools for sample inventory-based projects; we invite you to share your results using this forum.

[Countin'Trees]

Thanks for the info - the Minneapolis Case Study report is particularly valuable!

We are having our first planning meeting next week to launch our street tree sample survey project. We have our sample size but have not had time to pretest the software on a handheld.

Are there examples that we could obtain of what Stratum puts out as it's report?

Are there sample databases of other inventory projects that can be shared?

Is there a list of cities/contacts who have successfully used Stratum for a complete inventory or a sample survey?

Thanks

[StormyJerry]

We don't like to see posts go unanswered, and apologize that yours languished for so long. No users have picked this up, and we ourselves have been totally engaged in preparing and delivering a 4-day workshop.

There are other members of the i-Tree Team who can respond to your questions in detail later, but for now please look at the two STRATUM projects on this forum listed under Project Descriptions. One of them has a link to a report that should get you started.

Hope this helps.

[smaco]

Also, the US Forest Service has published many "Municipal Forest Resource Analyses" that are based on STRATUM outputs. See the following link for more information: http://www.fs.fed.us/psw/programs/cufr/ ... ProjID=151

[Countin'Trees]

It appears that for iTree formatted (Using the PDA utility) data; one must use TIGER and enter the TIGER TLID.

We would prefer to not use TIGER. We have many streets that cross parcels (private streets) that we would like to exclude and we would like generate maps that show the public right of way. TIGER will not allow us to perform these tasks. Is there a work around to this issue?

If I am entering a sample survey, what report actually produces the estimated total number of trees? What information is required to obtain the stocking level report?