Nocturnal satellite imagery and the analysis of spatial change

Nocturnal satellite images may offer an interesting tool to generate socio-economically relevant data and to analyse the evolution of space, e.g. cities or rural areas, and how spatial units interact over time. So far, the major purpose of using night satellite images for economic analysis has been the search for proxies for production and population density in countries with insufficient and unreliable data infrastructure. This essentially applies to less developed countries where weak data infrastructure is often part of overall underdeveloped administrative capacities. Error variance of light emission is constant over space and independent from error in official statistics. In industrialised countries official socio-economic data are deemed sufficiently reliable, a the reason why night satellite analysis has been more of relevance for developing countries. However, this differentiation only holds for purposes to derive proxies for production or population data. In fact, there is also reason to use this tool in the observation of (spatial) economic patterns and trends in the more industrialised countries. Spatial distribution of rural areas, urban agglomerations, border areas or other spatial categories are to be mentioned. More importantly, underlying patterns of spatial heterogeneity, such as Zipf's law, or spatial dependence (change of spatial autocorrelation over time) can be made visible - without distortion implied by (changing) administrative boundaries. The images analysed in this paper are satellite images of the National Oceanic and Atmospheric Administration (NOAA). Global nighttime lights imagery data are collected by the Defense Meteorological Satellite Program Operational Line Scanner (the DMSP-OLS). The night imagery data have a spatial resolution of 30 arc-seconds and are recorded between 75°N and 65°S. We explore whether time series of nocturnal satellite imagery and the application of adequate image analysis software, such as ImageJ (in some cases to be complemented by further statistics software), opens a useful perspective for the analysis of spatial change. ImageJ is a widely used open source image analysis software that had been developed at the US National Institutes of Health since the end of the 1990s supported by a large international community of contributors. Its primary purpose has been to process images in medical and biological sciences where has obtained a de facto standard for image analysis. However, as the synopsis of literature shows, the software has a broader range of applications - beyond microscopy and medical imaging - such as astronomy, environmental analyses, earth sciences including remote sensing, material sciences, viticulture archaeology and others. This paper is an essay with first ideas for discussion; the approach is explorative-methodological rather than one putting an empirical focus on a defined research item.