Arboriculture & Urban Forestry 46(1): January 2020 Calculations of Other Metrics Also calculated for each lot were basal area (ba, sum of cross section area of trees at dbh in m2 divided by lot area in hectares), trees per hectare (tph), and qua- dratic mean diameter for an estimate of average tree size (qmd = [200 Marshall 2000; Kershaw et al. 2017). In addition, car- bon (assumed to be 50% of biomass) was calculated (Chojnacky et al. 2014) for interpreting results. These metrics were developed for traditional forestry, so caution should be exercised when using them in urban forests in ways beyond the scope of this study; for example, modifications might be needed in our use of the entire lot (including impervious surfaces) as the basis for calculations. ], in cm)(Curtis and Analysis The preliminary calculations above and statistical analyses were conducted using SAS/STAT® soft- ware version 9.4 (SAS Institute Inc., Cary, North Car- olina, USA) and spatial analysis was done with ArcMap™ software version 10.3.1 (Esri, Redlands, California, USA). SAS/Graph® was used to create statistical graphics. Statistical testing assumed a sig- nificance level of 0.05. Because each lot was consid- ered a sample unit, tree data were summed to per-hectare lot-scale for analysis (Table 1). Objective 1: Lot-scale Canopy Cover Assessment An estimate of canopy cover 10 years after redevel- opment for each lot was obtained by assuming a gen- eral canopy growth curve as a function of time since redevelopment (period, or years, between Time1 and Time2), “indexed” to growth on each specific lot: lnc- growth = ß0 + ß1lnperiod + ß0 = cover2 – cover1 index, where: c-growth , period = years since redevelopment, -statistic = 0.98, n = 21). Then 10-year adjusted data were obtained by solving the equation for 10 years after redevelopment (lnperiod = ln10 = 2.3026). Cover at year 10 for each lot was then calculated by adding canopy cover at redevelopment (cover1 ) to the 10-year growth prediction from the equation. We hypothesized that this calculated canopy cover at year 10 would be greater than 20%; a one-sided t-test was used to test this hypothesis (H0 = 20, HA > 20). After statistical testing was conducted, statistical graphics were created to help interpret the entire ©2020 International Society of Arboriculture index = c-growth / period, ln = natural log (Table 1). This canopy growth curve was first fit to data (lnc- growth = -0.5633 + 0.9898 × lnperiod + 0.5833 × index; R2 17 sample distribution: canopy cover was easily com- pared to other metrics, and the graphs provided the perspective of “years since redevelopment” for each lot or inventory period. Objective 2: Lot-scale Model Development Modeling was done in two parts with per-hectare scale data (Table 1): (1) canopy cover predictions were developed from basal area (Mitchell and Popo- vich 1997), and (2) average basal area growth was estimated so that canopy growth could be projected. The correlation between nonoverlapping canopy cover at Time2 and basal area at Time2 was the basis for modeling canopy cover predictions from basal area, but the model also included quadratic mean diameter (qmd) at Time2 and an indicator variable to separate growth rates for planted trees from those for preserved trees. The model was fit using robust regression (regression modification where effects of outliers minimized; SAS Institute Inc. 2016). To model basal area growth, we defined average annual growth as the difference between Time1 and Time2 basal area divided by years between Time1 and Time2 (or period in Table 1). We separated data into four major categories—planted and preserved trees within deciduous and evergreen (hardwood and conifer) classes—to group basal area for these cate- gories into similar ranges. A model was then fit to each category to estimate an average annual basal area growth rate from Time1 basal area. Robust regression and log transformations were used to esti- mate parameters; regression was aimed at prediction only, so our primary interest was evaluating the model with respect to data fit rather than other regression diagnostics. The following were computed from Table 2 equa- tions in order to examine the overall statistical fit of data modeling: 1. Average annual basal area growth (bag) of each lot was estimated from equations for the respective categories ( bagdpr ˄ ˄ ˄ ˄ ,bagepr ,bagdpl ,bagepl , for decidu- ous preserved, evergreen preserved, deciduous planted, and evergreen planted, respectively). 2. Basal area at Time2 was estimated from basal area growth model results multiplied by the period between Time1 and Time2 in years (1 to 18) and added to Time1 basal area (for example, ( badpr2 ˄ = [bagdpr • ˄ period] + badpr1 ˄ ).
January 2020
Title Name |
Pages |
Delete |
Url |
Empty |
Search Text Block
Page #page_num
#doc_title
Hi $receivername|$receiveremail,
$sendername|$senderemail wrote these comments for you:
$message
$sendername|$senderemail would like for you to view the following digital edition.
Please click on the page below to be directed to the digital edition:
$thumbnail$pagenum
$link$pagenum
Your form submission was a success. You will be contacted by Washington Gas with follow-up information regarding your request.
This process might take longer please wait