Urban area is an essential parameter in urban surface modeling and greatly affects the results of meteorological models. Rapid industrialization and urbanization have increased significantly in the Pearl River Delta (PRD) region, especially, Guangzhou and Shenzhen have experienced marvelous changes in city construction in the past decade. Therefore, there is an urgent need to evaluate the type and area of new land use/land cover (LULC) in Guangdong Province, which is an essential part of the smallest research domain (Domain 4) of the Weather Research and Forecasting Models (WRF) applied in Southern China. This thesis investigates a method for classifying the urban area in the PRD based on building extraction from Google Maps.

The method described in this thesis is a two-step pro...[ Read more ]

Urban area is an essential parameter in urban surface modeling and greatly affects the results of meteorological models. Rapid industrialization and urbanization have increased significantly in the Pearl River Delta (PRD) region, especially, Guangzhou and Shenzhen have experienced marvelous changes in city construction in the past decade. Therefore, there is an urgent need to evaluate the type and area of new land use/land cover (LULC) in Guangdong Province, which is an essential part of the smallest research domain (Domain 4) of the Weather Research and Forecasting Models (WRF) applied in Southern China. This thesis investigates a method for classifying the urban area in the PRD based on building extraction from Google Maps.

The method described in this thesis is a two-step procedure: building detection and urban area classification. A method of automatic building detection that can separate individual buildings from surrounding features is presented. The process is realized in a hierarchical strategy, in which roads and buildings are sequentially detected. Major research efforts are made on the development of an image processing technique that accurately detects buildings from a set of colors of Google Maps. A method of urban area calculation is developed based on the area of constructed building, which is called the top plan area. The first stage is to find a correlation between the top plan area and the urban area with a set of sample data. The second stage is to implement this correlation formula in PRD region. The computed urban area is compared with official data or images in several cities with the PRD region, the method was shown to be able to detect the buildings and the accuracy of the method is no lower than 80% for modeling purpose.