Return to Record Keeping and Analysis | Return to comparison of image analysis and dot grids         .Guidelines for Developing and Evaluating Tree Ordinances

Evaluation example: CITYgreen software for ArcView GIS


One goal of a community's urban forest management strategy might be to maximize the benefits that trees provide, such as air pollutant removal, stormwater runoff reduction, and/or energy conservation. If this is the case, quantification of tree-related benefits might be useful for assessing current conditions or evaluating the results of ordinances or other management actions. Various researchers have developed formulas that can be used to estimate the magnitude of benefits related to tree canopy. Economic values associated with these ecosystem services can be calculated in some cases. For instance, energy savings can be converted into economic terms by multiplying the difference in energy usage by local utility rates. However, a complete analysis should also account for the value associated with avoided costs, such as avoided air pollution emissions associated with reduced power consumption. Researchers continue to develop and refine the formulas and parameters used to calculate tree-related benefits.

American Forests, a nonprofit citizen's conservation organization that focuses on trees and forests, has developed software designed to simplify the process of quantifying certain benefits that tree canopy provides. CITYgreen software is an application that uses embedded formulas and parameters to calculate tree benefits from information that is stored in a GIS. CITYgreen software is technically a third-party GIS extension of ArcView® GIS software available from ESRI, Inc. Extensions such as CITYgreen can only be used in conjunction with the base GIS program. Extensions consist of databases, scripts, and other objects that expand the capabilities of the base GIS and/or provide shortcuts that simplify various tasks or calculations.

We tested CITYgreen 3.0 software running under ArcView® GIS 3.2 to determine how this product might be used for evaluating progress toward urban forestry goals. CITYgreen and ArcView software was generously contributed by American Forests and ESRI, Inc, respectively, and provided a significant portion of the matching funding required for the NUCFAC cost-share grant that funded this website. At the time of this review (October 2001), later releases of both CITYgreen and ArcView® GIS have become available, but we have not had the opportunity to test them.

CITYgreen 3.0 actually provides two different ArcView extensions: CITYgreen Local Analysis and CITYgreen Regional Analysis. The Local Analysis extension can be used to calculate the economic value of a particular neighborhood greenspace, providing measurements of trees' contributions to stormwater runoff reduction, energy conservation through shading, air pollutant removal (ozone, sulfur dioxide, nitrogen dioxide, particulates 10 microns or less in size [PM10], and carbon monoxide removal), carbon storage and sequestration, and urban wildlife habitat. As the name implies, the Regional Analysis extension is used for regional analyses that cover a wider area. The Regional Analysis extension includes tools for detecting change in vegetation cover using certain types of satellite data and for calculating tree benefits on the scale of a large watershed.

A good working knowledge of how to use ArcView is a prerequisite for successfully using CITYgreen 3.0. The software interfaces for CITYgreen 3.0 and ArcView 3.2 are not completely intuitive, so unless you use these applications frequently, you may need to refer frequently to the manuals for help. In general, the manual provided with CITYgreen 3.0 was fairly helpful and made the extension relatively easy to use. However, in some instances windows shown in the manual did not correspond with onscreen windows. No online support for CITYgreen software was available at the American Forests web site at the time of this review.

Local Analysis Extension

Digitizing canopy and other features

To use CITYgreen 3.0 Local Analysis extension functions, the user must prepare a detailed schematic drawing and conduct a detailed tree inventory. Local Analysis calculations require information on the percent of land area covered by vegetation, water, structures, and impervious surfaces. These quantities are determined from GIS layers (or themes in ArcView parlance) that the user digitizes from a base aerial image or plan of a project site. CITYgreen provides special tools which are used to "heads up" digitize the base map and create a schematic drawing of the project site.

In CITYgreen, tree canopy is digitized by manually superimposing green circles over tree canopies shown in the base aerial photo. One also has the option of digitizing groups of trees as a single polygon, but several of the analyses that CITYgreen provides cannot be run on groups of trees. To test the canopy digitizing function, we digitized the trees in the same aerial photo shown on the page Comparison of image analysis and dot grids for calculating tree canopy cover. We used only the single tree method (superimposed circles) in the example image shown below. This method is fairly fast and directly produces a GIS layer that can be manipulated and analyzed. Each tree is given a unique identifier number by the program.

Although this method was fairly fast and simple, we noted several disadvantages. A certain amount of error is introduced when superimposing circles over the trees, especially because the image of the tree canopy is obscured as the circle is drawn. Canopy circles cannot overlap the edge of the project area or they will not be counted by CITYgreen. This causes problems if sizable numbers of trees are present along the perimeter of the project area. Furthermore, the user cannot modify the sizes of the circles that represent tree canopy through direct data entry. Canopy area and perimeter measurements for each tree in the tree attribute database are calculated from the digitized image when the analysis functions are run.

Canopy cover calculated from the digitized schematic (below right) was 17.38%. Canopy cover on the same image calculated using either image analysis or dot grid counts was about 21%. Unless tree canopies are very distinct and generally well separated, the CITYgreen method of digitizing canopy is likely to be subject to more error, especially between different evaluators, than these other methods. CITYgreen's alternative method for digitizing tree canopy essentially involves drawing polygons around tree canopy. This method may be more accurate if done carefully, but would be excruciatingly slow on an image such as we used for the example. Furthermore, CITYgreen treats groups of trees digitized as polygons as individual trees, which causes problems in assigning tree attributes.

Aerial photo Photo with digitized canopy

 

Other elements (i.e., layers or themes) that the user needs to digitize include buildings, impervious surfaces, grasslands, and water bodies. An example of a fully digitized image from a suburban neighborhood, with buildings and impervious surfaces as well as trees, can be found at the American Forests web site. Note that CITYgreen 3.0 does not calculate the area present in each of the land cover classes after digitizing is complete. It requires you to collect and enter field inventory data before it will run these calculations.

Modeling tree growth

The Local Analysis extension includes a tree growth model that allows a user to estimate the future benefits provided by a population of trees as they increase in size. In the tree growth model, stem diameter (DBH) growth is based on classification of trees as slow ( 0.1" DBH/year), medium (0.25" DBH/year), or fast growing (0.5" DBH/year). Height growth is modeled in an analogous fashion. CITYgreen models canopy growth by multiplying the the expected growth in DBH by a canopy growth factor specific for each species that was derived from the relationship between DBH and canopy spread measured in the field.

We tested the growth model on the digitized photo shown above right. For purposes of the analysis, we set the species code to "oak" because the western oak species present in the photo (Q. lobata, Q. wislizeni, and Q. douglasii) were not in CITYgreen's master species database. We set the height class to 15-35 ft and health to "good" for all 428 trees in the digitized image. We selected 10 years as the growth increment to model. The figure below shows the original canopy sizes (bright green) superimposed over the projected canopy after 10 years of growth (dark green). Based on our experience, it was clear that the default canopy growth rate for oak in the master database was greater than would actually occur with these species at this site. The user can modify the canopy growth rate in the master database to make it appropriate to local conditions, but local data would be needed to develop realistic numbers.

Trees near the edge of the project pose problems when modeling tree growth. Tree canopy that grows beyond the edge of the project is excluded from any of the environmental calculations that use canopy area. One alternative is to edit the tree theme by moving trees entirely within the site boundaries. This introduces some error into the calculations and alters the actual coordinates of the moved trees, which may be undesirable. Another alternative involves using the "split polygon" tool to manually edit out the portions of the trees that extend outside of the boundary. However, edited trees become irregular polygons and excluded from certain analyses.

Tree growth modeling

Calculating benefits provided by trees

Most but not all of CITYgreen's environmental benefits analyses require data in addition to the digitized schematic. The stormwater runoff reduction and air quality analyses only use the canopy size information digitized in the schematic, so these analyses can be run using only the data digitized from an aerial photo or site plan. However, CITYgreen requires that certain tree attribute fields be filled in before any analyses can be run. These include tree species code, trunk diameter, tree height class (<15 ft, 15-35 ft, >35 ft) and health class (the program uses a 1 to 5 scale). Hence, to run the stormwater and air quality analyses without these tree data, dummy data must be inserted into these fields in the database. This is most easily accomplished by opening the DBASE-format attribute table using a spreadsheet program and filling in the necessary fields.

The carbon sequestration analysis can also be run without field inventory data if the average diameter class of the tree population is entered for all the trees in the database. The remaining analyses (energy savings, growth models, and wildlife habitat) require information that needs to be collected through ground survey methods. Once the digitized site schematic is complete, it can be printed out for use in ground truthing and as an aid in collecting detailed field inventory data for the site. Datasheets containing the fields and codes used by CITYgreen are in the appendix of the user manual and are also provided on the CITYgreen CD-ROM. The manual also indicates which fields are required to run the various analyses. The datasheets provide a number of additional fields that might be useful for other purposes, and a user can add additional fields as desired. Hence, the tree database could be used as a tree inventory, although it would not necessarily have many of the specialized functions found in dedicated tree inventory software.

Using dummy data in required fields, we were able to run the analyses for carbon storage and sequestration and air pollution removal. The stormwater runoff analysis failed to run as described in the manual, which did not match exactly with the program. Nonetheless, we were able to model stormwater runoff with our data using the Off-site menu. The Off-site menu allows the user to model stormwater runoff by entering the required information directly in an input window that does not require linkage to a digitized schematic. The off-site menu for stormwater runoff modeling is especially useful for comparing the impacts of different scenarios involving, for instance, varying levels of canopy cover.

Regional Analysis extension

The Regional Analysis extension of CITYgreen includes a satellite data classification program and a watershed analysis program based on the 1992 National Resources Inventory (NRI) developed by the US Department of Agriculture. These two programs are independent and do not interact with each other in any way.

Satellite data classification program

We did not test the satellite data classification functions of CITYgreen. According to the user manual, the program uses a Normalized Difference Vegetation Index (NDVI) to classify pixels in a satellite image as either vegetation or nonvegetation, using information from the red and near infrared spectral bands. This means that tree canopy cannot be distinguished from other vegetation, such as grass. A statistical program automatically analyzes the satellite data and reports the vegetative cover in percent cover categories (0%, 1-5%, 6-20%, 21-40%, 41-60%, and >60%). A new image is created which displays the vegetation cover classes. Printing functions allow the image to be printed along with legends and a frequency distribution of vegetation cover classes. If satellite photos of the same region are available for different years, a change detection analysis can be run which will produce a new image showing vegetation cover change in the region.

Satellite data files must be in a band interleaved by pixel (.bip) format to function properly. An ArcView world transformation file must be associated with the .bip file prior to processing. The satellite data must be projected in UTM with map units in meters. Landsat MSS, TM, SPOT, and EOSAT data can be stored or converted to this format.

Watershed analysis program

The regional watershed analysis program in the Regional Analysis extension uses the 1992 NRI data to calculate the environmental benefits that vegetation provides in terms of carbon storage and sequestration, air pollution mitigation, soil erosion reduction, and stormwater runoff reduction for an entire watershed. We tested the program for our local watershed, the Lower Sacramento (CA), and found that the program functioned as described in the manual. The results of running the analysis functions are shown in the presentation below, which was prepared using an ArcView template preprogrammed by CITYgreen. Other layouts could be create using ArcView tools. Labels in the figure are county names, some of which did not print. The Lower Sacramento watershed is highlighted in yellow.

Regional analysis output

The program allows the user to model how changes in land cover statistics will affect air pollution removal by trees, carbon statistics, and stormwater runoff statistics. CITYgreen includes a template for printing results of the model scenario in an attractive format. Effects of changing land cover statistics on soil erosion cannot be modeled.

Final considerations

Modeling environmental benefits provided by trees and other vegetation is an ongoing topic of research. CITYgreen provides such modeling capabilities to non-experts, which is both its strength and weakness. By allowing routine calculations of tree-related environmental benefits, CITYgreen provides a way to take these benefits into account in community forest planning on both a local and regional scale. Nonetheless, the cookbook approach used by CITYgreen has the potential to result in flawed or unrealistic analyses in the hands of users that do not appreciate the intricacies and uncertainties involved in these analyses.

For instance, USDA Forest Service research indicates that for urban trees, the amount of carbon expended in maintenance activities over the life of a tree may exceed that stored by the tree. Energy conservation may be the main way in which trees reduce atmospheric carbon, but avoided carbon emissions associated with reduced energy use is not modeled in CITYgreen. There is also considerable uncertainty about the amount of carbon stored by forests. See for example: http://www.royalsoc.ac.uk/policy/carbonsinks_sum.pdf (the full report is at http://www.royalsoc.ac.uk/files/statfiles/document-150.pdf). Hence, the carbon storage and sequestration numbers generated by CITYgreen should be interpreted with caution. As a second example, CITYgreen's air pollution removal analysis is based on research conducted by Dr. David Nowak, of the USDA Forest Service, and is based on data from eight large US cities (Atlanta, Austin, Baltimore, Boston, Milwaukee, New York, Philadelphia, and Seattle). Pollutant levels in these cities may not be representative of levels present at a given project site or watershed, leading to erroneous (generally excessive) estimates of air pollution removal benefits.

The CITYgreen manual provides technical details on the analysis programs used to calculate environmental benefits and provides a list of references used to develop the analyses. In some cases, other programs exist that can be used to directly calculate some of the quantities calculated in CITYgreen. For example, the stormwater runoff analysis is based on the Urban Hydrology for Small Watersheds model (also called Technical Release 55 or TR-55) developed by the US Natural Resources Conservation Service (NRCS). A Windows-based version of the current TR-55 model is available for free download from NRCS at http://www.wcc.nrcs.usda.gov/water/quality/common/tr55/tr55-beta.html.


Return to Record Keeping and Analysis | Tree ordinance web site map |Return to comparison of image analysis and dot grids

ISA home page   | Submit comments or suggestions