| dc.description.abstract | Air pollution is problematic for a densely populated city like Ho Chi Minh City (HCMC).
Thanks to commercial and governmental organization’s air monitoring systems on the ground, citizens
now have access to AQI level in the metropolitan area of HCMC in real time. In research, international
studies have looked into the prospect of incorporating satellite data into monitoring air pollution, in
particular, PM2.5 pollution, using simple linear regression (Xie et al., 2015; Xu & Chang, 2020), or
more advanced GTW (He & Huang, 2018) or machine learning (Wang et al., 2019) methods. This
powerful combination between a great amount of matching data both from the sky and on Earth helps
researchers estimate PM2.5 pollution accurately, at high (Sentinel: 10m, Landsat: 30m, Terra-Aqua:
1km) spatial resolution over a wide area, and over a long period of time (Terra-Aqua, since 2000). In
this thesis, I explored Landsat-8 and Sentinel-2 atmospheric scattering, ground PAM Air PM2.5
concentration of HCMC in the dry monsoonal season (Jan – May) of 2020, and these variables’
relation, in search for a reliable method to estimate ground PM2.5 concentration from satellite data.
The method I chose is univariate linear regression. The result shows that level of confidence is low
(maximum R2 ~ 33%) for each sub dataset segmented from the initial dataset of Landsat-8, Sentinel-2,
and PAM Air PM2.5. In this thesis, I tried to detail each step of the process so that the result can be
reproduced for re-examination and further improvement in future work. | en_US |
