Mapping urban growth using Soil and Vegetation Index and Landsat Data: the Milan (Italy) city area case study

Remote sensing based on mid-resolution multi-spectral data has proven a powerful tool in urban areas study. This work introduces a novel methodology based on spectral indices ratios for mapping urban changes in terms of impervious surface expansion. At the methodological core, the Soil and Vegetation Index (SVI), a spectral index aimed at discriminating urban from non-urban land cover, has been utilized over Landsat TM-ETM+ satellite data. As a case study, the approach was applied to a multi-temporal dataset, with the aim of mapping the urban growth of Milan, Italy, during 20 years (1984-2003). The multistep processing framework is composed of: SVI values derivation and normalization, multi-temporal SVI ratios thresholding for identifying urban growth area, and multiscale segmentation of urban change maps produced. Results analysis showed the feasibility of the approach and reliability of urban change maps derived, which reached a value of Overall Accuracy up to 80%, while multi-scale assessment of results revealed the 25 pixels segmentation scale as the optimal one for urban change detection using Landsat data over the study area.