30DayMapChallenge

2023 Maps

I am doing the #30DayMapChallenge with work colleagues Geoffrey and Kerry. I will only be doing every third day.

A list of 30 maps to make for the 30DayMapChallenge

Day 1 - Points

Gathered data with Jupyter notebook by scraping website data into CSV.
In QGIS normalized data, used “Random Points in Polygons” tool to generate points.
Stylized map in QGIS and exported as SVG.
In InkScape I added text, did formatting/layout.
Would have liked to take data into R and do a bit of statistical analysis to see which, if any of these rates of blank ballots are significant.
EDIT: There’s an error in the last map. It’s just wrong. See day 13 for more information.

Three maps showing blank ballot counts in Luxembourg. The latter two are normalized by population and then also area - the third map is invalid due to a calculation error

Updated map

There was an error in my earlier map (above).

The above map was made in QGIS while the one below was created in R. The problem with QGIS is that I didn’t track what calculations I did to create the values.

In R, the code is explicit and maintained, making a review or repair much easier. The lesson is, when possible, always do it in R if it’s anything other than pure cartography.

The map below shows that this is really not the best way to represent this type of data. See the Day 13 map for a better representation of similar data.

Three maps showing blank ballot counts in Luxembourg. The latter two are normalized by population and then also area. It's hard to see any clear communal difference

Day 4 - A bad map

Used the osmdata library to get the features for the area from OSM
Map basic map in R. I should have gotten the symbology nicer there first…
Used InkScape to clean up data, organize, and then style a bit. Would have liked to do more, but… time.
The ‘bad’ aspect is that I seem to be promoting the ‘bad’ restaurants, or simply showing ‘bad’ restaurants.

A map of City of London showing the location of restaurants with 0-2 stars for hygiene rating

Used my GPS tracks from bus commute
Imported into QGIS and converted to paths
Set scale of the three maps in layout manager and exported
Added text and tweaked layout in InkScape

Three maps that look like spaghetti but are actual bus tracks over time

Day 10 - North America

I’m really impressed by the map by @xcaeag and want to apply the method. This is taking a bit longer.
I’m mapping one fire in North America, likely the largest of 2023. It’s in Northern Canada at the BC, Alberta, and Northwest Territories border.
I’m using hexagons, as per the day 9 map method I’m replicating (and using code from).
I automated the generation of svg tiles - don’t have time to do the art, but these can help do that.
For the SVG tiles, when exported from QGIS as SVG, then merged into one shape, the direction of the paths may matter. Also, QGIS duplicates each shape twice - so joining requires some separation.
My adapted script allows a specific hexagon size, this helps in the SVG marker settings in QGIS, as you can double the value (as my reference svg tiles have double the size of the contained hexagon).
This was much more complex than planned, and resulted in something that could have been done without the whole tiling to begin with. A lot was learned about the method though!
More details

A map of the fire coverage of the Hossitl creek wildfire in northern canada. Map uses hexagons to show fire extent, because hexagons are the bestagons.

Day 13 - Choropleth

Blank and invalid ballots in Luxembourg’s October 2023 legislative election.

I wanted to do more with my Day 1 map… look at the statistics. I discovered an error in my methodology… need to fix Day 1 at some point.

Benefit of using R is reproducibility. Not exactly sure what I did wrong in Day 1 with QGIS.

Map of Luxembourg communes with colours representing values. A graph in the bottom right shows that a few communes may have unexpectedly high rates of blank or invalid ballots

Day 16 - Oceania

Made with R, rayshader, InkScape

Learned how to use the R package rayshader. Big install but impressive. Need a higher quality DEM next time.

Rendering of Ululu - A 2km long rock rising up in the middle of flat lands

Day 19 - 5 minute map

Made with QGIS, Inkscape

Map of commute to school

Day 22 - North is not always up

Made with R, giscoR, and Inkscape.

Map showing migration of magnetic 'North' from Canada towards Russia

Day 25 - Antarctica

Made with R/giscoR, selected area in QGIS, export as SVG, and styled in InkScape.

A monchrome (blue) map of the James Ross Island Group in Antarctica

Day 28 - Is this a chart or a map?

Methodology

Get a DEM/DSM for an area of interst.

In QGIS draw a line with rounded turns, such as through a valley. Use the Digitize with curve tool (Ctrl-Shift-G) for curves.

Surface elevation model of Luxembourg City with a line following the valley bottom

Import the scripts to the QGIS Processing Toolbox (PT) by clicking the Python icon in the PT and select “Add Script to Toolbox…”.

The use the script Perpendicular Line Creator (perpendicular_lines_creation.py) to generate a bunch of perpendicular lines at a specific interval (e.g. 10m) and length (e.g. 200.5m). The extra 0.5m is to prevent edge cases later.

Parallel lines to the earlier line

Now have the DEM/DSM be sampled along the perpendicular lines and create a tab separated file. The script output needs some work but functions fine - just needs a bit of cleaning the output. I’ll fix it when I have time.

Use the Raster Profile Extractor (RSE) script (extract_profiles.py) provided.

Take the data into R, clean it up, and plot it.

You can also use RSE on satellite or aerial photography to sample the colour at the sampled locations.

Here are the alternatives possible based on whether you use a DSM, DTM, or orthophoto.

Four graphs of elevation lines perpendicular to a path

You can also use a simpler path - but it’s not quite the same. Here’s a line, rather than a curve, coming up the Bock (a rock formation in Luxembourg city).

Parallel lines up Luxembourg City's Bock