Blog #3 – Don't Stop Believing (It's All About the Journey)
Welcome back and happy #mapmonday! Today we’re going to be taking another look at whether there are any cities in Canada that have settled away from major water sources. But before I get into it, I want to emphasize that the answer itself is secondary to the journey we are going to take to get there. Using GIS to answer questions 9 times out of 10 will involve a very meandering path, with multiple options to answer questions, and realizations that your original plan was, in fact, bunk. That definitely happened while I was writing this post, and it’s happened to me many times before. But it’s my hope that pointing that out will encourage anyone who’s struggled with making maps to keep trying, and to get good at googling. Seriously, learning the language of GIS and how to do effective searches will do wonders to find the help you need to answer your spatial questions.
So far, we have only been using the global dataset of cities, or “populated places” from Natural Earth. We’re going to continue with that dataset today, keeping in mind that there are many smaller Canadian cities not included in this global dataset. Last time we found that a number of cities that were further than 20 km away from rivers in the Canadian river shapefile were either close to a coast, or close to a larger waterbody not included in the river shapefile. This week, we are going to take another look into this query by expanding it to include other types of waterbodies.
I considered a couple of options for doing this, one of which was combining the line and polygon shapefiles by converting the line shapefile to a polygon shapefile, but what do I do about any overlap? What names do I include? Should I combine polygons that share the same name? What if two rivers on opposite sides of the country have the same name? It got a little complex, plus, I’d have to include the coastal areas, and I was a little stressed about it to be honest. But when I sat down to get this blog started today, I had a realization: I don’t have to combine all the shapefiles, I can just use the Near tool to compare the cities to all datasets at once!
We used the Near tool last week to see how many Canadian cities were more than 20 km away from the river shapefile, the one that didn’t include the lakes and large rivers. What the tool does is measure the distance from each point (the cities in this case) to the nearest feature (the lakes, rivers, and coasts), which can be a point, a line, or a polygon. We can select multiple shapefiles for our near features (see figure). The measured distance is then recorded in the attribute table, along with the name of the closest feature.
Using GIS gives us so many different options to answer the same questions, there are many paths to take to find your answer. Because of this, we can try a number of different things. My first instinct here was overcomplicating the situation, which would have ultimately led me to choosing a path that has more steps. It’s still possible that the method I’m choosing could be simplified further! But for now, I know what I need to do, and it’s easy enough to get started.
At first, I tried to run the Near tool multiple times, once for each dataset, but it turns out, GIS doesn’t need that. However, when I tried to run Near with all three datasets, it crashed! I thought maybe it was an issue with overwriting data in the table, but eventually I figured out that the Canada shapefile was causing GIS to crash because all of the cities are inside of it (woops). I converted the Canada shapefile to a line feature using the Feature to Line tool so I could use with all three features (if, for some reason, you’re following along at home, dissolve the Canadian border line features into one feature to avoid more frustration).
After this debacle, I ran the tool with the both the river (CADRivers_Dis) and lake (CanadaSW) shapefiles so I could visualize if any remaining cities were near a coast. There were 8 cities outside of the 20 km range: Port-Menier, Hopedale, Coral Harbour, Grise Fiord, Port Burwell, Igloolik, Resolute, and Nain, all of which are on the coasts of the Northwestern Passages, the Gulf of St. Lawrence, or the Labrador Sea.
This would lead to the conclusion that there are no Canadian cities in the Natural Earth database that settled 20 km away from surface water, which isn’t surprising at all really, but I wanted to take a closer look at these results.
I made a buffer around the cities to visualize what 10, 20, and 30 km would look like and I found that the set distance was effectively working as a radius instead of a diameter (woops). Meaning that the assumption was that cities could be 40 km wide (definitely a minority there). So, I decided to try it again with a 10 km distance and a 5 km distance.
This time, instead of setting a distance in the Near tool, I left it blank. Turns out there’s a simpler way to do this test: instead of having to re-run the Near tool over and over again, we can just calculate distance with the Near tool and use Select by Attribute to find cities that are a set distance away.
When I took a look at the distances, there were no cities that had a recorded Near distance of more than 20 km, which we found before. Narrowing down the search to 5 km and 10 km, there were three cities that were more than 10 km away from a water source (left image), and seven cities more than 5 km away from a water source (right image). These cities are (from closest to furthest to a water source): Bella Bella (BC), Wetaskiwin (AB), Owen Sound (ON), Camrose (AB), Port Hope Simpson (NL), Cartwright (NL), and Steinbach (MB).
Interesting! I’m curious about these cities and their historical significance. Do they use groundwater as their main drinking water source? Or do they have water transport systems in place? Maybe we can answer those questions another time, for now, I want to make a map! We have the distance to surface water as a data column now, so we can use it to define our symbology.
The redder the dot, the closer it is to a surface water source, and we can see the light coloured cities we highlighted earlier which are more than 5 km or 10 km from a lake, a river, or a coast. Next week we can look into why this is, and maybe find a more comprehensive dataset of all the cities in Canada to do a more thorough examination now that we have the methodology down.
Getting to this answer involved a lot of changed plans and re-worked ideas, but that is all part of the fun. Working with GIS has taught me to think more creatively about problem solving, although it may have given me some pretty big headaches in the process. I never stop believing that I can get to the answer somehow, and I’ve learned that when trying to conquer GIS, it’s not about the answer to the question, it’s about the journey.