Automatic mapping of urban areas from Landsat data using impervious surface fraction algorithm
Monday, 15 December 2014
Urbanization is a result of aggregation of people in urban areas that can help advance socioeconomic development and pull out people from the poverty line. However, if not monitored well, it can also lead to loss of farmlands, natural forests as well as to societal impacts including burgeoning growth of slums, pollution, and crime. Thus, spatiotemporal information that shapes the urbanization is thus critical to the process of urban planning. The overall objective of this study is to develop an impervious surface fraction algorithm (ISFA) for automatically mapping urban areas from Landsat data. We processed the data for 1986, 2001 and 2014 to trace the multi-decadal spatiotemporal change of Honduran capital city using a three-step procedure: (1) data pre-processing to perform image normalization as well as to produce the difference in the values (DVSS) between the simple ratio (SR) of green and shortwave bands and the soil adjust vegetation index (SAVI), (2) quantification of urban areas using ISFA, and (3) accuracy assessment of mapping results using the ground reference data constructed using land-cover maps and FORMOSAT-2 imagery. The mapping accuracy assessment was performed for 2001 and 2014 by comparing with the ground reference data indicated satisfactory results with the overall accuracies and Kappa coefficients generally higher than 90% and 0.8, respectively. When examining the urbanization between these years, it could be observed that the urban area was significantly expanded from 1986 to 2014, mainly driven by two factors of rapid population growth and socioeconomic development. This study eventually leads to a realization of the merit of using ISFA for multi-decadal monitoring of the urbanization of Honduran capital city from Landsat data. Results from this research can be used by urban planners as a general indicator to quantify urban change and environmental impacts. The methods were thus transferable to monitor urban growth in cities and their peri areas around the world.