Quick Links

Milwaukee Ash Tree Identification



Client

City of Milwaukee Department of Public Works, Environmental Services, Forestry Section

Goal

  • Utilize cutting edge Hyperspectral and LiDAR technologies to map the locations of White and Green Ash tree species throughout the city of Milwaukee, Wisconsin
  • Generate landcover maps from hyperspectral-derived color-infrared images
  • Conduct an urban tree canopy (UTC) study
  • Generate CITYgreen reports for each aldermanic district in Milwaukee

Product and Process

This project applied the most advanced geospatial technology, including high-resolution remotely sensed Hyperspectral and LiDAR imagery, in conjunction with GIS analytical applications, to develop new tools needed for improved species mapping, risk assessment, forest health monitoring, rapid early detection, and management of EAB.

For this project, airborne hyperspectral data represented a unique capability to support the classification of green and white ash trees over a large area while addressing some important shortfalls of large-scale inventories and analysis projects.  The collection and analysis of airborne spectral imagery from a single integrated HSI and LiDAR platform ensured that the data collected for the campaign maintained a coherent spatial and temporal framework. Simultaneously, the use of hyperspectral data for tree classification allowed the project to determine and maintain a consistent, measureable accuracy level across the AOI and avoid subjectivity that can result from ground inventories conducted over a large time span.

RFP Mapping LLC, NCDC Imaging, SRA International, and ASD inc. partnered with the Services Forestry Department to conduct a field spectral data collection campaign focused on gathering field spectra of the ash species of interest and potential tree and vegetation species that may be within the AOI. The field data collection campaign consisted of a structured process designed to collect sample spectra of species of interest and surrounding environmental and spatial information in order to support the development of a robust methodology to detect and characterize green and white ash within the Milwaukee area.


Graph showing ground-collected spectral signatures of 8 common tree species


SRA adapted an established hyperspectral analysis process to detect and characterize white and green ash. The analysis methodology consisted of two primary tasks – selection of spectral signatures and selection of analysis algorithm and approach.

The analysis team utilized the field collected spectra to develop three signatures for exploitation – white ash, green ash, and common ash (representing a commonality across ash species to improve the detection of some ash variants such as purple ash). The development of these signatures followed an interactive process during which the field collected spectra were compared with spectra extracted from the airborne hyperspectral data.

Using sophisticated GIS analytical tools, NCDC converted the dataset from SRA into a vector data format and intersected the resulting file with the tree points from the LiDAR extraction, which were buffered by fifteen feet. Analysts were then able to select the tree points contained within the tree point buffer and attribute them with a “high” or “low” level of confidence that any particular point was an ash tree based on proximity and size. As a general rule, any tree polygon measuring less than 15 square feet, within the buffered distance would be attributed with a “low” confidence level. Because the tree canopy layer was attributed with the square-foot area of each polygon, geoprocessing steps can be applied to calculate current and possible urban forest canopy cover for the entire City and for each Aldermanic district as well as having integrated the ash species maps to improve targeted inspections, outreach efforts, urban forest mitigation and influence public policy.




*Check back for a video interview with Milwaukee Forestry Services Manager David Sivyer