Context: we analyze economic changes in Colombian municipalities between 1993 and 2020.

After having downloaded nighttime images from NOAA National Centers for Environmental Information (NCEI), and install ArcGIS or any other GIS software in our computer, we proceed now to upload nighttime files for our analysis.

NOAA provide images covering the whole world which can be difficult to manage in terms of size. For that reason, we will only upload the images corresponding to our study are, i.e. Colombia. To do that, make sure you have uploaded first the map of the study are, here Colombia (see previous post)

In ArcMap, we go to Windows—Search and one window with search engine will show up in the right side.

In order to extract only the images corresponding to Colombia, we search the tool “extract by mask”. There you will find further details on this tool. Extracting all the nighttime light files we should have stored previously in our folder is the first step to build our dataset of light intensity, i.e. our economic indicator.

Once we click on “extract by Mask” this window will appear asking for input raster, which is the night light file stored somewhere in your computer, in my case, in the folder I stored all the years “nightime light time series DMSP-OLS”. Then, input raster or feature mask data. This is the “mask” properly, and the way to say the software you only want one part of the whole image, in this case, the boundaries of Colombia, so we choose “COL_adm2”, the shapefile we previous open in ArcMap. Finally, in output raster, this is the file that will store all the data from the nighttime light series, the file that will contain our dataset for our analysis.

After clicking OK, we will be able to visualize Colombia image at night

We will extract as many files as years we want to analyse. In our case, we uploaded from 1993 to 2013, which as visible in the “table of content window”.

But before making that dataset look like a classic dataset we use in Spss or similar statistical softwares, we need to convert all the nighttime light files into a different format. These files are in format raster. In computer graphics and digital photography, a raster graphics or bitmap image is a dot matrix data structure that represents a generally rectangular grid of pixels, viewable via a bitmapped display, paper, or other display medium.

Importantly, NOAA assign a value from 0 to 63 to each of the pixels, where 63 means the maximum light intensity. However, in practice, we cannot see the value of each of the pixels we need for our analysis, so that we need to convert this raster into “Points”, meaning that every single pixel you see in the above image will be converted in a point with a given value, this time associated to a table of content.

We go again to window—-search and look for “raster to point”. We have two options, selecting “raster to point” and convert all the raster one by one into points or simply “extract value to points”, which will allow us to extract all the values from all the raster and then convert them into points.

Option 1

Option 2

Following the option 2, we select the file where we previously created “dataset” (in the image basedatosrecurse) and add all the rasters by clicking +. We click ok and wait until the program is able to perform the transformation.

The software will now show a new layer in the Table of content window “basedatosrecurse” (or dataset as labelled above) Clicking right bottom with the cursor on this layer, we “open attribute table” and see all the data with a classic dataset used in Spss. Every record correspond with one tiny point in the map (former pixel in raster format) and have assign a value from 0-63 for each of the years, where 63 indicates greater economic growth. You can see the dataset I created for Colombia here in my ResearchGate profile https://www.researchgate.net/publication/344282484_Shedding_light_on_the_local_resource_curse_in_Colombian_coal

We are almost ready to perform our advance economic analysis. Before, we need to link each of the points with the municipalities. See our next posts.