2008 - Ahmedebad: Using RS and GIS for Indicators for Urban Transport Ecological Footprint Analysis
MSc thesis submitted by Ron Dalumpines to the International Institute for Geo-information Science and Earth Observation in the Netherlands.
The research explores the extraction of urban form/land use information in developing indicators to support TEF analysis using RS and GIS. Remotely sensed imagery provides a global information resource that when compared to traditional methods of data collection has the ability to provide data of an entire area, of areas that are difficult to access, at a greater frequency in acquiring data, reusable for different projects and can be cost efficient. GIS supports handling of spatial data from remotely sensed imagery and integrates it with other images and ancillary data from different sources. Recent RS and GIS applications can handle various spatial analyses and other data manipulation techniques considered useful for data mining, such as indicator extraction and quantification. In this study, urban RS plays a key role in providing thematic classifications (i.e., residential, commercial, institutional, and industrial classes) based on IRS-P6 satellite imagery. Perpixel classification methods, supervised and unsupervised, grey-level co-occurrence matrix texture measures and spatial metrics are explored in the extraction of four urban land uses for indicator quantification. The utility of freely available high-resolution Google Earth images supported by global positioning systems (GPS) are also explored in the process. The utilization of RS and GIS applications is further illustrated in the extraction and quantification of TEF-related indicators, namely, density, proximity, trip distance estimate, and land-use mix.
Example results for the case study city of Ahmedabad in India provide preliminary insights into the challenges in deriving indicators from RS imagery for transport ecological footprint. This study shows that supervised classification method (overall accuracy = 54.87%, Kappa = 0.0706) have limited capability in extracting residential, commercial, institutional, and industrial building classes. Incorporation of optimal texture measures (Kappa = 0.1137) has a potential in improving per-pixel classification results. Applications of basic statistics, categorical analysis, and spatial metrics in quantifying RS-derived indicators demonstrate inconclusive (R2 = 0.007) links between urban form and TEF proxy, i.e. total number of trips. Hence, these methods are considered limited in assessing the transport ecological footprint of the city. For more insightful results, it is recommended that better image classification methods together with more sophisticated, model-enhanced indicators be employed. RS and GIS are highly applicable in this kind of endeavour.
Keywords: Remote sensing; GIS; Transport ecological footprint; Urban form; Image classification; GPS; Google Earth; IRS-P6; Spatial analysis; Spatial metrics
|Last Updated Date: