Date of Award
Master of Public Health (MPH)
INTRODUCTION: Identifying the amount of vegetation in small geographic plays an important role in understand the contribution of the built environment to Heat Stress Illness (HSI) in urban areas. Vegetation is often identified using remotely sensed satellite imagery, using the Normalized Difference Vegetation Index (NDVI). However, it is possible that small areas, such as census tracts or block groups, may be sensitive to the resolution of the sensor used for NDVI classification.
AIM: The aim of this study is to determine if the sensor resolution of remotely sensed data produces significantly different NDVI classifications for census tracts and block groups in the Atlanta area for 2011. In addition, we examined the role of landcover classification in understanding these differences.
METHODS: Using the 2010 geographic designations for census tracts (n=142) and block groups (n=358) in the Atlanta area, mean NDVI values were calculated and compared. The imagery used was from the Landsat5 (30m resolution) and QuickBird (2.44m) satellites taken in June 2011. Proportions of landcover classifications were calculated using the 2011 National Land Cover Database and compared to NDVI mean differences.
RESULTS: QuickBird classifications were significantly higher than the Landsat5 classifications at both the census tract and block group level (p value). We found correlation for NDVI difference with standard deviation of the high resolution NDVI values (r(356)=-0.61, p
DISCUSSION: High-resolution imagery produced higher NDVI values for census tracts and block groups for the city of Atlanta in 2011. Our analysis suggests that High-resolution imagery may be beneficial in improving accuracy of identifying urban neighborhoods at higher risk of HSI.
Vinson, Daniel, "The Effect of Sensor Resolution on Detection of Vegetation in the City of Atlanta, Georgia in June 2011." Thesis, Georgia State University, 2018.
Available for download on Wednesday, May 01, 2019