Site icon Javier Cha

Visualizing the Korean Peninsula using ASTER DEM data

I am as much a skeptic of the digital humanities as I am a practitioner. In particular, I have a lot of reservations about applying advanced GIS methods on plain or spotty samples that are better served by manual plotting or choropleth maps. This is especially the case with data sets extracted from medieval and early modern sources.

Nonetheless, in my DH courses I need to expose students to all aspects of humanities computing, and that includes GIS. A lot of quantitative historical sources from the nineteenth century forward provide spatial information with the granularity necessary for meaningful spatial statistical analysis. As a graduate student at Harvard, I did learn the basics of computational spatial analysis at a two-week GIS training workshop organized by the Center for Geographic Analysis. (I can’t believe that was so far back in the past, in 2009.) I also worked closely with C. Scott Walker, an incredibly knowledgeable and skillful digital cartographer at Harvard Map Collection, for several sets of maps we produced for the Early Korea publication series.

Now that I am expanding the scope of my research interests to areas far beyond medieval Korean studies, it makes sense to create some useful high-resolution basemaps. Back in 2009, Scott introduced me to two most versatile and highly precise data sets for creating basemaps: GSHHS and ASTER DEM. GSHHS is a set of vector data for global shorelines. I will get to that in a future blog post. For the time being, I decided to learn how to handle ASTER DEM.

ASTER DEM provides ultra-high resolution raster data of the planet’s land surface at an impressive 30-meter resolution. To my knowledge (which is very limited), it is presently the most detailed elevation data publicly available. The data is based on satellite images collected by Japan’s Ministry of Economy, Trade, and Industry (METI) and has been made available for general public use jointly by METI and NASA.

For one of the Early Korea volumes, Scott made a gorgeous map of archaeological sites near Kyôngju, an ancient Korean capital, using a raster basemap based on ASTER DEM. I wanted to see if I could create a high-resolution raster basemap for the whole of northeast Asia, so I gave it a go. Since 2009 or 2010, computers have become much faster and GIS software much more advanced, yes? Yes, but I underestimated the sheer size of uncompressed TIFF files that contain such high level of detail. At first, I tried to stitch together a very large collection of ASTER tiles. That didn’t go so well. So I decided to limit the scope to the Korean peninsula. It worked. It also created an enormous 3 GB GeoTIFF file. To provide some perspective, the DEM image provided by China Historical GIS v4 is 65MB in size and has a resolution of 1km. In my guesstimate, the ASTER DEM tiles that cover the whole of today’s PRC territory will result in a GeoTIFF file in the region of 30 to 40GB.

In any case, below you will find a gallery with four high-resolution images (8000+ pixels or roughly 1200 dpi on A4 or US letter paper) that I generated using the open-source software QGIS. Notice that these are not maps with scale, legend, and north indicator; they are simply demonstrations of the sort of physiographic visualizations that the ASTER DEM data set makes possible. The first image shows the Korean peninsula after stitching together the pertinent tiles: it is the “raw” data, so-to-speak, and processing is limited to the application of pseudocolors. The subsequent close-up images have hillshading applied: notice the incredible level of elevation detail that can be seen.

As a follow-up to this exercise (which has been mainly for my own learning), I will visualize a part of the world other than Korea and put together a tutorial on how I generated these images using open-source software and the publicly-available ASTER DEM.

Use the following plug-in to zoom in and check the level of digital elevation detail of each map tile. Please note that some images are 20 to 30 MB in size and therefore may require some time to load fully.

Exit mobile version