Monitoring two decades of urbanization in the Poyang Lake area, China through spectral unmixing

There is an increasing need to understand the dynamics in urbanization not only temporally but also spatially for the improvement of urban environments. In spite of an enormous number of previous studies in urban remote sensing applications, only a few studies have been conducted on the techniques in the quantification, qualification, and visualization of the changes in time-series urban land cover fractions (LCFs) derived through spectral unmixing. To examine the urbanization process in four major cities around the Poyang Lake area in Jiangxi Province, China – Nanchang, Jingdezhen, Yingtan, and Poyang – using a time-series Landsat-5 TM dataset in 1987, 1993, 1999, 2004, and 2009, we investigated: (1) the approach to the derivation of LCFs in urban areas using multi-temporal remotely-sensed data set; and (2) the approach to the summarization and cartographic manipulation of the changes in time-series LCFs. To account for the complex spectral confusion among different land cover materials in built-up areas, the Multiple Endmember Spectral Mixture Analysis (MESMA) was used for unmixing the pixels. The Land Cover Change Intensity (LCCI) was proposed to derive the average daily change rate in terms of the area within a pixel for the land cover classes of green vegetation, non-photosynthetic vegetation and soil, and built-up areas between two consecutive TM observation dates. The dominant LCCI (DLCCI) was proposed to determine in which period the urban areas were developed most rapidly and how intense the urbanization process was in each pixel of the time-series LCCI maps. Our results showed that MESMA could accurately model the pixels in urban areas with complex spectral confusion of different land cover materials. The comparison of derived land cover fractions with socioeconomic statistics disclosed the strong positive correlation between built-up fractions and urban population as well as gross GDP and GDPs in secondary and tertiary industries. LCCI and DLCCI revealed two mechanisms of urbanization, which are new land developments and redevelopments of built-up areas. Consequently, we found that the four cities around the Poyang Lake were urbanized through different mechanisms.

► MESMA could accurately model the pixels in urban areas.
► Strong positive correlation was found between BU LCFs and socioeconomic statistics.
► Proposed LCCI and DLCCI could distinguish land development and urban redevelopment.
► The four cities around the Poyang Lake were urbanized through different mechanisms.