The ground survey is one of the most basic methods for gathering urban forestry data. Ground surveys typically are used to gather the baseline data for most tree inventories. The ground surveys used in urban forestry are of two general types, commonly referred to as windshield surveys and foot surveys. Details of each type are discussed below. When resources are insufficient to conduct a complete inventory, a representative sample of the urban forest can often provide sufficient information for making management decisions and monitoring progress. Furthermore, when natural woodlands or forests are managed, as in parks and open spaces, a complete inventory is usually unnecessary and impractical.Sampling from Populations. To provide estimates of size and condition of street trees, researchers working in several cities in the eastern U.S. (Valentine et al 1978) arrived at the following recommendations, which can be used as a rough rule of thumb for planning ground surveys:
-Canopy dieback. This is a simple indicator of tree health. Either tally trees above and below a given cutoff value (e.g., dieback affecting more than 1/3 of the crown), or use 3 to 4 categories (e.g., low, moderate, severe, tree dead). If descriptors such as "low" or "severe" are to be used, it is necessary to establish specific criteria for each description (e.g. low=less than 20% of crown affected) to minimize differences that may arise between different evaluators. Photographs that illustrate the different classes are very useful to ensure uniformity between different evaluators and different years.
-Improper pruning practices. Topping and other poor pruning practices are especially obvious in winter after leaf fall.
-Prohibited practices, such as vandalism, or attaching signs or wires to trees.
-Specific disease and pest problems. If surveys are timed to coincide with periods when disease or insect pest problems are most obvious, it may be relatively easy to document the extent and incidence of the problem. For example, leafy mistletoe in deciduous trees is easily rated in the winter months, whereas branch dieback in alder caused by flatheaded borers is most obvious in summer.
-Tree type. Trees can be placed into relatively broad categories based on height or type (e.g., conifers, evergreen hardwoods, deciduous hardwoods) fairly readily. Also, the frequency of a single or a few distinctive tree species could be tallied. However, especially in areas where a wide variety of tree species are used, a complete tally of trees by species would be difficult or impossible to conduct from a moving vehicle.
-Trunk diameter. For many, though not all tree species, diameter serves as a useful indicator of tree age. Several broad classes of tree diameters (e.g. less than 6 inches, 6- 24 inches, greater than 24 inches) can be distinguished with enough accuracy to be used in a windshield survey.
-Planting site characteristics. Empty planting spaces, severe sidewalk displacement, and other obvious site characteristics can be tallied.
When detailed information in a number of different categories is to be collected, the survey should be conducted on foot. All of the examples listed above under the windshield survey could also be evaluated in a foot survey. Some data may be expressed as categories, as in the windshield survey, but it is also possible to take more detailed data and actual measurements rather than use generalizations and estimates. For example, stem diameter can be measured rather than estimated and trees can be identified to species. The type of planting space (for example grass, bare soil, depressed well, level well, raised planter) and size of the planting space can be identified. Tree condition, hazardous trees, hardscape damage, and site conditions can be inspected and evaluated more thoroughly in the foot survey than in the windshield survey.
If data are being collected for an inventory, such as a street tree inventory, data are typically collected for every tree in the area of interest. If forests or woodlands containing large numbers of trees are being evaluated, it is more efficient to sample the area rather than conduct a complete survey. Sampling may occur using plot-based or plotless techniques (e.g., point-centered quarter method). Plots may be arranged in various shapes and sizes, and plots of varying sizes are sometimes nested within each other. Plot area can be either fixed (e.g., circular 0.2 acre plots) or variable (e.g., prism-based plots). Plots may be permanent, which allows for direct observations of changes that occur over time. Given the wide variety of sampling methods available for measuring forest attributes, persons that specialize in forest survey methods (e.g., university forestry department faculty, forestry consultants, state and federal forestry staff) should be consulted before undertaking a forest survey.
Plot or tree locations can be noted directly on maps or aerial photos. Standard survey techniques can also be used to pinpoint tree locations. With the decreased cost and increased precision of GPS (Global Positioning System) technology, the use of hand-held GPS receivers provides another way to determine tree or plot coordinates in the field. However, GPS readings from low cost units are subject to several sources of error that can degrade the precision of location information. In particular, tree trunks, branches, and canopy can interfere with the reception of satellite signals needed to obtain coordinates. We have been able to achieve improved reception by using a high-gain external GPS antenna mounted on a mast that can be elevated at least part of the way into the canopy.
Some common measurements recorded in foot surveys are described below.
Canopy cover provided by individual trees can be estimated by measuring the maximum canopy diameter and a second diameter at a right angle to the first. Canopy area can then be calculated using the formula for the area of an ellipse, i.e.,
Area = pi * r1 *r2
where pi=3.14159, and r1 and r2 are the two radii (i.e., half the diameters). If tree canopies are symmetrical, a single diameter can be measured and the formula for the area of a circle (pi*r*r) is used. The total area covered by tree canopy can be divided by the area of the site to obtain percent canopy cover. This methodology works best for areas with nonoverlapping tree canopies, such as parking lots or other relatively open areas.
In areas with more complete or irregularly overlapping tree cover, other methods of estimating canopy cover are applicable. If data are being collected in individual fixed-area plots, ocular estimates of tree canopy cover may be adequate. A density scale (see the example under Photogrammetry and remote sensing) can be used to help calibrate different observers. Also, less error will be introduced if canopy is estimated in cover classes, such as the six-level scale discussed below.
Two similarly named instruments can also be used for measuring tree overstory canopy cover: the spherical densiometer and the densitometer. The two terms are sometimes interchanged, so either term may be used to describe either type of instrument. The spherical densiometer is used to measure canopy cover over a plot or other local area. An image of the canopy is reflected onto a gridded spherical mirror and the observer counts the number of points on the mirror that either contain or lack canopy cover. The number of points counted is then divided by the total number of points to calculate percentage. Several replicate measurements are needed to increase precision. Densiometer measurements are influenced by adjacent canopy height and tend to overestimate canopy cover because canopy is viewed at an increasingly oblique angle toward the edges of the mirror. Photos taken with a hemispheric or fisheye lens can be used in a similar fashion, except that canopy cover is evaluated on the images rather than directly in the field. Hemispheric photos have the same biases as spherical densiometer measurements. Bias can be reduced by using a smaller view angle (about a 10 degree arc), which reduces bias associated with oblique viewing angles.
The densitometer provides a point measure of canopy cover. The densitometer is a small sighting instrument with crosshairs and a bubble level that allows the observer to determine whether canopy is present directly overhead. This instrument is sometimes referred to as a moosehorn, and several variants exist. Since the densitometer measures canopy presence at a single point, multiple sample points must be measured to obtain a canopy cover estimate. Sample points can be spaced along a transects (see the example Measurement of canopy cover at the edge of pavement (CCEP)) or arranged in a grid pattern to obtain an estimate for a large area. Using a densitometer is directly analogous to using the dot grid method to estimate canopy cover from aerial imagery. Consequently, sample size considerations are the same as discussed for the dot grid method.
Tree trunk diameter at breast height (4.5 ft height if English units are used) is one of the most commonly measured tree size statistics. However, tree form, ground slope, and other factors can complicate this measurement. We have developed a Simplified guide to measuring DBH that discusses a number of these common issues.
There are many methods for measuring tree height. Tree height can be measured directly with a calibrated measuring pole or indirectly through trigonometric relationships by using a clinometer or a similar device. Many websites describe methods for measuring tree height. .
Evaluating tree condition is always a subjective enterprise, because such evaluations rely on visual assessments made by observers. The simplest scales rate the condition of living trees as good, fair, or poor. If more detail is needed, various aspects of tree condition are independently rated. Certain ratings (e.g., canopy thinning or live crown ratio, decay ratings) provide information about chronic health problems, whereas others (e.g., current season foliar symptoms) reflect recent health impacts. Quantitative rating scales (discussed below) can simplify assessments and reduce variability between different observers.
The USDA Forest Service Inventory and Analysis program has developed detailed standardized methods for rating tree condition and many other tree and plot factors. Illustrated manuals describing these methods in detail are available online. Detailed scales for evaluating tree health and condition developed by The Urban Forests Centre at the Faculty of Forestry, University of Toronto, which are part of the Neighbourwoods inventory program, may be available on the Internet. The Neighbourwoods program is designed to minimize bias among different surveyors. This website includes scales and in some cases photographs for evaluating the following conditions:
Some of the factors listed above relate directly to a tree's hazard rating. An illustrated guide to hazardous trees is available online at the USDA Forest Service St. Paul Field Office web site. ISA publishes a widely-used guide titled "A Photographic Guide to The Evaluation of Hazard Trees in Urban Areas, 2nd edition". This publication can be ordered from ISA.
Conflicts that develop between trees and infrastructure are often evaluated in ground surveys. The proximity of overhead wires, buildings or other structures, other trees, traffic signs, and sidewalks and curbs can all require management actions to maintain public health and safety or tree health. Distances between trees and various hardscape elements can assessed by measuring distances directly (using tape measures, distance measuring wheels, or rangefinders) or can be rated qualitatively based on visual inspection (not a problem, potential/future problem, current problem). If damage or conflicts are present, the nature and extent of the problem can also be noted and prioritized for corrective action.
Various types of tree assessments do not lend themselves to direct measurement but can be estimated visually. For instance canopy dieback, an important tree health parameter is difficult to measure directly but the percentage of the canopy affected by dieback can be estimated by a trained observer. Other assessments, such as canopy cover, can be assessed using reasonably precise methods, but the amount of time and effort required may not be justified based on the use of the data. In such cases, ocular estimates may be used even though more precise methods are available.
As noted above, visual rating scales can be developed for many of the assessments that are made in ground surveys. Qualitative rating scales can be quite objective if only the presence or absence of a characteristic is noted (e.g., presence/absence of leafy mistletoe). Subjective qualitative scales (e.g., rating mistletoe infection as light, moderate, or heavy) are also commonly used, but it can be difficult to obtain consistent ratings from multiple observers when subjective scales are used. However, such scales can be useful if qualitative categories are augmented with more quantitative explanations (e.g., light mistletoe rating=less than n infections). Photo keys that illustrate different qualitative rating classes can also help make qualitative ratings more objective and repeatable among different evaluators.
Quantitative rating scales are also commonly used. Scales are used to simplify the estimation of quantities such as counts or percents. Different types of scales ay be appropriate for different types of ratings. For instance, when estimating percent cover in small circular area (e.g., within the dripline of a tree), a scale using 25% increments (0-25%, 26-50%, etc.) is typically easy to use and can be rated consistently. For estimating plot canopy cover, canopy dieback, or other quantities that can vary across a wide percentage range, the following 0 to 6 scale is useful:
|0||none / not present|
|1||more than 0 but less than 2.5%|
|2||2.5% to less than 20%|
|3||20% to less than 50%|
|4||50% to less than 80%|
|5||80% to less than 97.5%|
|6||97.5% or more|
Note that the classes in this scale are not uniform in size and are largest near 50% and smaller near 0 and 100%. Various studies have shown that observers are able to estimate percentages (such as percent cover) that are close to 0 or 100% with greater accuracy than they can estimate percentages near 50%. This makes sense intuitively. For instance, it is easier to distinguish between 2% and 12% cover than it is to distinguish between 45% and 55% cover, even though the absolute difference is 10% cover in either case. The scale above is similar to the Daubenmire (1959) scale except that the class edges have been modified so that the midpoints of the scale increments are equally spaced in an arcsine transformed scale. Percentage data is typically binomially distributed and the arcsine transformation (arcsine of the square root of the percentage expressed as a decimal) is used before this type of data is analyzed using standard parametric statistical tests. By using a pre-transformed scale (Little and Hills 1972), ratings can be statistically analyzed without further transformation.