Last Blog!

Research & Work

Because this is my last blog, I would like to reiterate the research I have done over the past eight weeks and describe its significance. 

The Heat Island Phenomenon - the heat island phenomenon occurs when highly dense built-up areas trap heat, making urban areas warmer than surrounding suburban and rural areas. Impervious ground cover like concrete or pavement prevents thermal energy from being absorbed into the soil. Bodies of water and vegetated areas lower air temperatures through heat of vaporization. According to the Environmental Protection Agency, "Heat islands can affect communities by increasing summertime peak energy demand, air conditioning costs, air pollution and greenhouse gas emissions, heat-related illness and mortality, and water quality" ( The network of cities committed to addressing climate change related issues like the heat island effect are known as C40 cities. The World Urban Database and Access Portal Tools (WUDAPT) aims to "gather and disseminate information [that describes aspects of the form and function of cities at a detailed spatial resolution] in a consistent manner for urban areas worldwide" ( WUDAPT uses characteristics like building height and density, vegetation type and density, and land cover type to categorize different regions of cities into Local Climate Zones. Ultimately, they aim to produce Local Climate Zone Maps of all C40 Cities, and possibly even the entire planet.

Training Areas - The first step in the LCZ mapping process is to select 10-20 training samples. Samples should be as diverse from each other as possible so that the generated map will consider varying types of land coverings and built-ups.

SAGA GIS - Once training samples are collected, LANDSAT Satellite images are loaded onto SAGA. These images show what type of buildings or land cover exist in each pixel. The more images used, the more accurate the LCZ map will be. SAGA uses the training samples (uploaded as a KMZ file from Google Earth) along with LANDSAT images to produce a colored LCZ map of the entire region based on how each LCZ is defined.

Confidence Report - Once the final LCZ map is generated on SAGA, a confidence report is made to check the accuracy of the map. Sample areas (separate from training areas) are selected after the map is generated. Once 3-5 samples are selected through random sampling, they are uploaded to ArcMap. The accuracy of the map is determined by the ratio of pixels that are correctly colored based on their LCZ classification to the total number of pixels in the selected sample area. These ratios are inputted into a confusion matrix where the overall accuracy and kappa coefficient are calculated. 

Expanding LCZ Maps - The WUDAPT team at CUHK is beginning to collaborate with a team of computer scientists from Carnegie Mellon University. The team, xSeer, developed a software called EVA that can process millions of data points at once and produces interactive diagrams. If the WUDAPT process can be implemented into EVA software, large scale maps of the entire United States, or even planet, could be generated. A worldwide LCZ map would allow researchers to compare global climate zones to local air temperature, as well as other factors including energy use, air quality, and evapotranspiration. Pairing these technologies has the potential to quantify the heat island effect which would allow architects, engineers, and city planners to consider the environmental and health impacts correlated to building density and design. I have been assigned the task of collaborating with the Carnegie Mellon team and relaying the WUDAPT process to them so they can figure out possible ways to implement the process into their software. 

My tasks for the next week and a half - I will be researching ways to upload kmz, geotiff, and grids files onto MATLAB. When I figure out an effective method to do this, I will perform regression analyses of the relationship between local climate zones and evapotranspiration. I would like to research this topic in particular because as an environmental engineer, I have an interest in the hydrologic cycle and how it might be effected or altered by urban density. I came up with the idea to research this topic when browsing the EPA website and reading that heat islands can affect water quality by raising its temperature, and therefore rendering it less able to store oxygen. 


Last Friday night I went on a 15 kilometer bike ride with my co workers, Margaret and Mona. We rented bikes on campus at University station and took a bike route around the entire bay to a reservoir in Tai Mek Tuk which is near Tai Po. It was so hot when we left but the views along the way were absolutely breath taking. We passed by parks, skyscrapers, hidden temples, and villages. We ended at the reservoir and rode our bikes half way across the dam to watch the sunset over the islands. Afterwards, we got Thai food in a local village.

The following day I went to Stanely beach with some friends I made in the program. We went to two different beaches, walked around Stanely Market, and had amazing food.

Next weekend I plan to hike to the geological park in Sai Kung Country Park. It apparently houses massive amounts of columnar basalt which is something I have wanted to see in person for a very long time!