Fine-Scale Urban Tree Change Detection in Worcester, Massachusetts
Course: Remote Sensing
Date: December 2018 Authors:Rachel Corcoran-Adams and Elizabeth Lohr Background: As urban greening initiatives expand throughout the country, it is becoming increasingly important that these programs measure their success. In his 2017 paper on the topic of mediating Worcester’s disposition to the Surface Urban Heat Island (SUHI) Effect, Arthur Elmes (2017) illustrates the importance of an increased urban tree canopy. The SUHI Effect is defined as increased temperatures in cities relative to rural areas (Elmes, 2017). After the loss of much of the tree canopy in certain Worcester neighborhoods as a result of the 2008 Asian Long-Horned Beetle outbreak, Worcester, Massachusetts’ urban canopy has been under siege more than ever (Elmes, 2017). This research analyzes the change in juvenile tree canopy cover in Worcester, Massachusetts’ Greendale and Burncoat neighborhoods by conducting a Mahalanobis classification. In addition, it allows us to obtain new knowledge working with high resolution satellite imagery. |
Data:
Acquisition Date |
Source |
Sensor |
Resolution |
Projection |
Bands |
September 7, 2015 |
Arthur Elmes |
Vector (N/A) |
Vector (N/A) |
NAD_1983_201 1_StatePlane_Ma ssachusetts_Main land_FIPS_2001 |
None |
June 19, 2018 |
Digital Globe (MAXAR) |
Worldview |
1m |
WGS84_UTM_Z one_19N |
Band 2 (blue, 448-510 nm) Band 3 (green, 518-586 nm) Band 5 (red, 632-692 nm) and 7 (NIR1, 772-890) |
Methods:
Outcome: There are differences between the two neighborhoods in tree cover over this time period. The areas of huge tree gain in the northern corners of the image are due to the fact that there was no tree cover data for 2015 beyond the city limits. As a result, the areas outside the city that we classified tree cover data for in 2018 are mischaracterized as tree gain, because there was no tree cover there in 2015. Additionally, while high-resolution imagery offers a multitude of spatial benefits, it also is susceptible to classification challenges, which leads to our discussion of future classification endeavors. In the course of our project we ran multiple iterations of the Mahalanobis classification after increasing the number of training sites, targeting areas that had been misclassified. Instead of increasing the accuracy, as we expected, the classification changed each time but not in any systematic way.
|