Arboriculture & Urban Forestry 42(6): November 2016 detect urban trees at a large scale and differentiate different tree species from one another. In contrast to RADAR data, LiDAR data on urban forest have been widely used and adopted (Forzieri et al. 2009; Nicholas et al. 2012; Oshio et al. 2012; Zhang and Qiu 2012; Zhu et al. 2012; Adeline et al. 2013; Nie- meyer et al. 2013; Oshio et al. 2013; Zhou 2013). Sung (2012) applied LiDAR data to assess the mean canopy height and percent canopy cover of an urban forest. According to the land develop- ment ordinance, the landowner should secure a permit to remove trees larger than 41 cm in diam- eter at breast height. Sung also utilized LiDAR data to establish the canopy height model (CHM) by calculating the difference between the digital elevation model (DEM) and the digital surface model (DSM) (only on tree canopy). The cells with values less than 1 m were not included in the analysis. Although the results demonstrated that LiDAR data are highly applicable and are recommended for use to detect the tree structure and evaluate the tree canopy height, the method mentioned (subtracting DEM from DSM) is not applicable in urban areas because the highest surfaces can sometimes be manmade materials, such as building roofs. When the uppermost sur- faces are not tree canopies, the difference between DSM and DEM cannot be used to solve for CHM. Many studies have been employed on different aspects of urban forests, including tree crown shape and structure (Oshio et al. 2012; Sung 2012; Oshio et al. 2013), tree detection and urban veg- etation mapping (Hofle et al. 2012; MacFaden et al. 2012; Yao and Wei 2013; Zhou 2013), tree posi- tion and plant density (Forzieri et al. 2009), and individual tree species detection (Nicholas et al. 2012; Zhang and Qiu 2012). Zhang and Qiu (2012) solved the limitation of LiDAR data by using hyperspectral imaging to distinguish more than 10 tree species, and LiDAR data to detect tree crowns only; thus, the accuracy of tree species detection was directly related to the resolution of the images. In another study, Nicholas et al. (2012) applied discrete Fourier transform on LiDAR data to distinguish five individual tree species. In other studies, the discrete point of LiDAR data has been shown to be more accurate than that of airborne waveform Lidar data when performing tree classi- fication (Reitberger et al. 2008; Hollaus et al. 2009; 403 Heinzel and Koch 2011). Nicholas et al. (2012), who were the first to investigate and compare the accuracy of tree species detection by using the dis- crete point of LiDAR data and waveform LiDAR data, reported that the overall accuracy increases by about 6.2% when waveform information was used (overall accuracy was about 79.2% when the discrete point data were used but increased to 85.4% when the waveform LiDAR data were used). Although the use of airborne remote-sensing data, such as LiDAR data, has many advantages, it also presents several limitations for urban land-cover classification; such limitations include processing and interpolating of point clouds into raster layers, which are time consuming and prone to misclas- sification (Zhou 2013). Appendix I shows the satel- lite and airborne sensors for urban forest studies. In conclusion, although the most suitable data that can be used to detect urban forests and dis- tinguish urban tree species are hyperspectral data, they are not recommended in urban forest studies because of their limitations, such as limited cov- erage, high volume, and cost. Thus, other satellite imaging techniques should be used. Moderate- resolution imaging techniques is unsuitable for urban forest detection because of its low spatial res- olution, which leads to mixed pixels. By contrast, high-resolution imaging techniques can be used to detect urban forests but cannot distinguish urban tree species unless an imaging technique with a higher resolution, such as WorldView-2, is used. The result of urban tree species classification using the traditional and new bands from the World- View-2 imaging technique has demonstrated the effectiveness of this method to distinguish urban tree species from one another, although some misclassification might occur because of spectral similarity. This limitation can be addressed by using miscellaneous information, such as spa- tial, texture, and color. The last data, which are almost used as an ancillary data, are LiDAR data. Although these data can extract urban features even in the shadows and at night, the processing and interpolating of point clouds into raster layers are time consuming and may lead to some misclas- sifications. Finally, this review has shown that the use of data from WorldView-2 provides the most efficient way to detect urban forests if cost, time, and accuracy are considered as research factors. ©2016 International Society of Arboriculture
November 2016
| 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
