Cities in pan Pacific

 

This is a summary of my PhD research.  

 

Cities strive for economic strength while recognize the necessity of being environmentally sustainable. The balance between economic development and the environment has been challenging particularly for cities in the pan Pacific region, which is seeing some of the most rapid urban growth rates.

My PhD project used time series of satellite images for 25 cities (Figure 1) around the pan Pacific region to investigate the dynamics between urban environment and economic development using Environment Kuznets Curve (EKC) theory as a lens, addressing how urban vegetation responds to urban nighttime brightness in 25 cities across the pan Pacific region.

 Figure 1. I selected 25 cities located around the pan Pacific regions with varying population sizes. 

Figure 1. I selected 25 cities located around the pan Pacific regions with varying population sizes. 

Two satellite derived variables, vegetation and nighttime lights,  were used as a proxy for urban environment and urban economy. . I first extracted urban vegetation using spectral mixture analysis (SMA ). Nighttime lights brightness was used to assess urban economic development and its relationship with census-derived variables.

Three models were fitted for each pixel (Figure 2). A linear model represents a monotonic and irreversible relationship. A quadratic model is similar to EKC theory. A cubic model indicates a more complex and undulating trend. The best model was selected based on Akaike information criterion. 

 Figure 3. We fit 3 models per each pixel. Water was excluded from the analsis. 

Figure 3. We fit 3 models per each pixel. Water was excluded from the analsis. 

The results indicated that the majority of the pixels follow a linear relationship (Figure 4). It means that vegetation is unlikely to recover once it's been damaged at least with the study period of this project (i.e. 1992-2012). This result is even more obvious in higher income cities. Pixels with statistically strong quadratic relationships between vegetation and brightness were less prevalent but more spatially clustered in comparison to those that expressed a linear relationship. 

 

 Figure 4. Histograms of r-square values for each of the three fitted models. (Note that I did not use r-square to select models. AIC was used as the primary ceriteria). 

Figure 4. Histograms of r-square values for each of the three fitted models. (Note that I did not use r-square to select models. AIC was used as the primary ceriteria). 

 

I have written two journal articles that give a more detailed review and descriptions of this project. 

They are currently under review. 

If you are interested in how I generated those vegetation time series. They can be found here, and here.