Alberto Federico OGAS MENDEZ , Tohoku University, Japan
Pei XUANDA, Tohoku University, Japan
Inequality and segregation are the main characteristics of urban problems in most cities in developing countries. The proliferation of informal settlements (Squatter Settlements) epitomizes that situation. This kind of settlement results from a previous squatting process outside the norms of the regulations of the urban planning code. It often presents many urban problems such as segregation, health concerns, environmental degradation, inadequate living conditions, and informality (UN-Habitat, 2003).
One of the main challenges for urban planners is integrating these settlements with the rest of the city. However, one of the obstacles in reaching that objective is the lack of an effective method for understanding and measuring informal settlements.
This research objective is to create an accessible and reliable framework of the locations, cadastral situations, and structural characteristics of the urban fabric inside the informal settlements. We expect that the results of this research will provide valuable information that helps in collaborating in the confection of a sustainable master plan for sensitive redevelopment in informal settlements.
The analysis method is through the Google Earth Engine (GEE) platform, a cloud-based geospatial analysis that allows us to visualize and analyze our image data. The data analysis is through a pixel-based Random Forest (RF) algorithm. It is generally immune to data noise and overfitting and is extremely useful in classifying remote sensing data (Teluguntla et al., 2018). It is considered one of the most widely used and accurate land cover classification algorithms (Millard & Richardson 2015). The RF can estimate the urban built-up area separately without roads, public places, green areas, etc. The decision trees will vote for each pixel to recognize the attributes and land use characteristics to create a cadastral classification of the informal settlements.
Mots clés : Informal settlements|Random Forest|Google Earth Engine|Remote sensing|GIS
A104081AO