I have used satellite images from NOAA (DMSP-OLS) to monitor and compare the distribution of luminosity of areas over time, e.g. urbanisation, peri-urban zones, borders etc. For me it is still the experimantal phase. For that purpose I have found the software ImageJ quite useful, particularly image correlation and - for urban development - analysis with concentric circles where one can also import the pixel data of the different circles into stata for further statistical analysis, e.g. Moran-I analysis. I would just like to learn and exchange. To my knowledge analysis of night satellite images have been so far only used to proxy missing income or population data in developing countries. Thanks in advance!
http://maps.ngdc.noaa.gov/viewers/dmsp_gcv4/
We used the night images to universally define or delineate cities or settlements in order to examine the universality of Zipf's law. It was found that Zipf's law is not only universal for city sizes, but applied to city numbers as well. That is, the number of cities in the first largest country is twice as many as that in the second largest country, three times as many as that in the third largest country, and so on.
https://www.researchgate.net/publication/260167281_Zipf's_Law_for_All_the_Natural_Cities_around_the_World
Article Zipf's Law for All the Natural Cities around the World
Thanks a lot! Your paper seems highly interesting for our work. I'll have a closer look at it. In the moment I still try to explore the potentials of ImageJ (software) for all those spatial evolution issues. Once I have something presentable on paper I'll let you share it. Many thanks again.
Thanks for your interesting question! The statistical analysis you indicated such as Moran-I analysis is essentially spatial dependence. What is missing in the literature of spatial statistics is spatial heterogeneity, or the scaling pattern of far more small things than large ones.
https://www.researchgate.net/publication/259875214_Geospatial_Analysis_Requires_a_Different_Way_of_Thinking_The_Problem_of_Spatial_Heterogeneity
Article Geospatial Analysis Requires a Different Way of Thinking: Th...
Many thanks for the second paper you sent. Meanwhile I tested the pixel distribution of light emission for Germany and could prove that the Zipf (rank-size) coefficient of the lights is much closer to -1 than that based on city population data. So I am more and more convinced of this approach of spatial analysis. By the way, one last question: Which software did you use to extract the pixel data from the NOAA images?
Good to hear that you are convinced of our approach of spatial analysis!
Note that there are different ways of extracting the so called natural cities. Relying on night images is just one way. The other ways includes using (1) street nodes and blocks, and (2) social media locations. This set of slides summarize the concept of natural cities, and importantly the underlying way of thinking on fractal geometry:
https://www.researchgate.net/publication/283723073_Fractal_Geometry_As_a_New_Paradigm_for_Urban_Structure_and_Dynamics_Natural_Cities_As_an_Example
I believe the approach of spatial analysis goes beyond the current spatial analysis, or spatial statistics, which is very much on spatial dependence. Our approach of spatial analysis focuses on spatial heterogeneity, or the fractal or scaling notion of far more small things than large ones.
As for software, it is ArcGIS. Please refer to this master thesis, in particular Appendices, where the detailed data processing is described: http://hig.diva-portal.org/smash/record.jsf?pid=diva2%3A692813&dswid=7890
Research Fractal Geometry As a New Paradigm for Urban Structure and D...
Dear Bin, it is highly impressive what can be done with such data. Admittedly my attempt to consider such satellite images has been so far a more practical one, as to test or confirm results from empirical social and economic research on rural and urban evolution, such as e.g. peri-urbanisation. I am also aware of the fractal dimension of space, such as the rank-size distribution of cities or developmental strengths and weaknesses. Implicitly this can be even found in policies, such as the cascade-like logic of the Europe2020 strategy, where the EU as such and every tiny sub-region is defined by the same developmental categories and challenges. The EU is just the aggregate of the variation at national down to local levels. Indeed, to my knowledge there is not much consideration of spatial heterogeneity in regional science. Some few remarks can be found in Brakman et al. 2009 (The new introduction to geographical economics, Cambridge: CUP). I do not have ArcGIS and no clear idea of its certainly powerful capacities. The original inspiration to use ImageJ came from my daughter who just works on her master thesis in biochemistry and the information on http://www.noao.edu/education/gsmt/lpmeasure. In the moment we are preparing a research application with other EU partners on rural & urban change. For this purpose I am preparing a technical paper on the image analysis approach. Your work will be duly considered with references. Once completed I'll let you know. Many thanks and best wishes, Rolf
- Badges
- Science topic
- Similar topics
- Filing