31:T459,${link(3)} ## *Day 3. Polygons. “[Antipodal Archipelago](https://observablehq.com/@visionscarto/antipodal-archipelago?collection=@visionscarto/30daymapchallenge)”* The topic was an opportunity to showcase two implementations of the *particularly difficult* problem of **clipping polygons**. The first one is based on the algorithm designed in 2009 by Francisco Martinez, Antonio Jesus Rueda and Francisco Ramon Feito, and [implemented](https://github.com/mfogel/polygon-clipping) by Mike Fogel, Alexander Milevski and Vladimir Ovsyannikov. I had spent a lot of time trying all sorts of implementations of Martinez-Rueda-Feito (and they were all failing in many corner cases), when I finally found this implementation, that seems quite robust. The second one is the implementation of spherical polygon clipping in the [d3-geo-polygon](https://github.com/d3/d3-geo-polygon) module, based on the pioneering work by Jason Davies. Probably the most complicated piece of software I’ve ever worked on—the kind that makes you feel you’ve disassembled a watch and have to reassemble it from all its tiny pieces.32:T4af,${link(22)} ## *Day 22. Boundaries. “[Guessing Czechoslovakia](https://observablehq.com/@visionscarto/guessing-czechoslovakia?collection=@visionscarto/30daymapchallenge)”* The **cartographic metaphor** can be very useful to describe a *field* (oops… spatial metaphor again) that is not spatial in nature. For example, people like to create *mind maps*, or *maps of the genome*, *maps of research networks*, &c. The tools we build to do that (**projections** known as “dimension reduction” algorithms, in many ways and forms…) are quite powerful in creating localities, neighborhoods, connections between neighborhoods. This creates a *map* with points (or nodes), streets and roads that connect them, localities that describe what is “near”, mountains that show what is “high-up” or “hard to reach”, valleys that concentrate flows. However, if we want the metaphor to hold, we have to test these tools against spatial data too, and see if the map we get out is intelligible, or if on the contrary the initial space is destroyed by that transformation. The Czechoslovakia example and data were given to me by Françoise Bahoken, as part of the *Tribute to Tobler* initiative.33:T5a6,${link(23)} ## *Day 23. GHSL data challenge. “[Growing and declining](https://observablehq.com/@visionscarto/ghsl-growing-and-declining?collection=@visionscarto/30daymapchallenge)”* The GHSL is a huge database, and this was an opportunity to showcase the fantastic work that my colleagues at Observable are doing to bring SQL to the notebook in the smoothest and most efficient ways. The map in itself shows cities that have been growing fast in the last 40 years vs. those that have been slightly declining… the geographic divide is absolutely striking. But you can edit the SQL queries and map something else entirely! The notebook also shows a preview of the Plot.carto plugin I’ve been writing for Observable Plot. A possible extension: convert the fields to the proper type before uploading, and keep only the fields I need, would make this notebook faster to load, and easier on the browser’s memory. It also makes me want to develop a canvas-based layer for Plot, so we can draw even more data points. Note: with the GHSL as with any database, the values given must be taken with some caution. For example, Naypyidaw, the administrative capital of Myanmar, is listed as having 167,000 inhabitants in 2015, vs 5.7 millions in 1975. It is quite possible that the number for P15 is mistaken by a degree of magnitude. To make a serious map, even with the best databases, there is always going to be a lengthy verification process.34:T467,--- ${link(30)} ## *Day 30. Meta. “[30 days and as many maps](https://observablehq.com/@visionscarto/30-days-and-as-many-maps?collection=@visionscarto/30daymapchallenge)”* Writing this, looking back on the last 30 days, I realize how much of the fun I have writing these notebooks depends on the existence and vibrance of a large community and on the activity of all the people (past and present) who did research, published it, compiled datasets, created software, created visualizations, and spread all kinds of enthusiasm; people who enjoy sharing, explaining what they do and encouraging others to try and make stuff, and get excited about new ways of seeing and representing spatial (or non-spatial) data. People I have mentioned above, and many many more. Shout-out to Mike Bostock, Jason Davies, Olivia Vane, Philippe Rekacewicz, Agnès Stienne, Jeffrey Heer, Nadieh Bremer, Claudio Esperança, Shirley Wu, Romain Vuillemot, Gabriel Peyré, Aurélien Tabard, David Cœurjolly, Julie Delon, Bruno Lévy, Irene de la Torre, Carmen Torrecillas, Anne-Christine Bronner and all the people of Transcarto…, among others!35:T739,The constraint of having to do one map per day forced me to limit the time I usually spend going down the rabbit-hole. Research —if I can use such a word— is never so much about getting answers than about finding new questions. When exploring some concept freely, you're always looking to go deeper, to find special cases, or new applications, and dig, and dig. Since I had to deliver one map every day, I couldn’t afford to do that, and it was… liberating. Many of these maps could certainly be improved a lot with a day (or a week) more work. But would they really be *so much* better? Many aspects of the algorithms or of the visuals could be made harder, better, faster—but instead of wanting to cover everything, I had to put that aside, and get something out. Besides time-boxing the work, another useful constrain was to avoid any *serious* subject. Doing a serious map means doing serious data work, and this always takes a lot of time. Making geometric games and cartographic tricks doesn’t carry that type of responsibility, and allows to just play and see where it goes. Many of these maps —in fact, almost all of them— are lacking a title and a legend. This is bad. The text compensates up to a point, but the maps should be readable independently of the text, and that's an additional layer of work I couldn't find the time to put in. Overall, I haven’t put enough effort in the finish, except for the Loir-et-Cher map, with help of my artist daughter Ines. When I compare to the works by [Maxene Graze](https://observablehq.com/@ambassadors/historical-map-day-24?collection=@ambassadors/30daymapchallenge), [Tereza Iofciu](https://observablehq.com/@ambassadors/30daymapchallenge-2021-9-monochrome), [Olivia Vane](https://observablehq.com/collection/@oliviafvane/watermarks) I know I have to focus more on this area.36:Taa5,Finally, a few questions: * **Why am I so fascinated by the early computer mapping algorithms?** Maybe another way of framing that question is to ask, *what have we lost* when geographic information systems (GIS) became dominant? Looking back at the research from the 1970’s and 80’s, it’s obvious that maps were not just the layering of tons of data on top of one another (if I can caricature what GIS does). Cartography was meant to be *transformative*, to show relations, movements, networks, structures of power. With the access we have now to fantastic new classes of algorithms, easy to plug in with data in notebooks that run instantly, there is a lot to invent, and we can iterate quickly, mix and match, try things out. We just need to do a bit of homework to learn and rediscover (and sometimes resuscitate) what the previous generation explored. * **What amounts of reprojection did I use?** (Thanks, Jo, for that question—I'm pretty sure my answer is completely off-topic, but here goes:) Like everybody I guess, I tend to bring everything into shapes I know how to deal with, and for which I have tools at the ready. This means I usually work in spherical coordinates (longitude, latitude), and a lot of the experiments focus in bringing planar algorithms to the Sphere. Sometimes it’s as easy as replacing the euclidian distance with the great-circle distance (WoS, SDF…), sometimes it’s totally hairy (d3-geo-voronoi). I also tend to prefer *graphs* to *grids*, an approach that brings a lot of simplification to many algorithms—though it means some optimizations are unusable. So if I can bring any type of cartographic data or problem to a graph living on the sphere, I'm usually happy. And when I don’t have a graph, I’d rather create one (with the delaunay of a random distribution of points) than use a rigid grid. *Then,* I will reproject the results to a flat map. * **What hacks gave great results for relatively little effort?** (Tom.) In general in this challenge I’ve tried to make the code as concise and readable as possible. But without question the Walk-on-Spheres notebook had the best bang for the buck: just three lines of code, and that opens a new way of painting a map with a “meaningful” method. Other notebooks that didn't need a lot of code for maximal effect were the Gabriel graph (*one* line of code gets you a very organic-looking graph), and the barycentric interpolation method. Often, using some kind of graph on a random distribution of points, with Voronoi and Dijkstra as the go-to algorithms (even when other algorithms are known to be faster), gives a nice perspective on “locality”. And optimal transport will soon join them in the toolbox.37:[{"id":"d0c8ba2cb61a102be717f123e3d66696da2e9117e1b9fa778051f53a90ccee5de479e839845a179d415e2700e8d5262bd391452f1dbef17edeaad653ebdf4556","href":"https://static.observableusercontent.com/files/d0c8ba2cb61a102be717f123e3d66696da2e9117e1b9fa778051f53a90ccee5de479e839845a179d415e2700e8d5262bd391452f1dbef17edeaad653ebdf4556","name":"collection.json","type":"application/json","lastModified":1638274638753.35,"size":47234},{"id":"8214ac75692e33852641af655f0fc87fbe4a1add40157037ccc17525842e6f15bfad9028a81143086deaaeae99507c9b1decf2f07da5af02c2d6ecf980697f42","href":"https://static.observableusercontent.com/files/8214ac75692e33852641af655f0fc87fbe4a1add40157037ccc17525842e6f15bfad9028a81143086deaaeae99507c9b1decf2f07da5af02c2d6ecf980697f42","name":"collection@1.json","type":"application/json","lastModified":1668544024179.088,"size":43554}] 2f:{"title":"Thirty days and as many maps","cells":[{"id":0,"value":"# Thirty days and as many maps\n\n## A look back on the #30DayMapChallenge, 2021.","pinned":false,"hidden":false,"mode":"md"},{"id":6,"value":"*Embarking on this challenge at the last minute was a bit stupid, but it felt like an opportunity to finish a few open projects, to try a few ideas, and to showcase a part of my usual toolbox.*","pinned":false,"hidden":false,"mode":"md"},{"id":15,"value":"$${link(1)}\n## *Day 1. Points. “[Dotted Oceans](https://observablehq.com/@visionscarto/dotted-oceans?collection=@visionscarto/30daymapchallenge)”*\n\nThis globe uses: a [backface **projection**](https://observablehq.com/@d3/projection-reflectx) that allows to “see through” the Earth, and a nice-looking [random point **distribution**](https://observablehq.com/@fil/2d-point-distributions). Although it’s a very simple notebook, it was quite successful.\n\n

","pinned":false,"hidden":false,"mode":"md"},{"id":38,"value":"$${link(2)}\n## *Day 2. Lines. “[World Cities Gabriel](https://observablehq.com/@visionscarto/world-cities-gabriel?collection=@visionscarto/30daymapchallenge)”*\n\nMethods that can build a “natural-looking” **graph** from a set of points are pretty satisfying, because they’re esthetically pleasing, and also because we can use them to simulate “real-life” mechanisms. I’m often using the [Urquhart graph](https://observablehq.com/@fil/the-urquhart-graph), planar or spherical. This time, I wanted to play with the [Gabriel graph](https://observablehq.com/@rpgove/constructing-a-gabriel-graph), but on the **sphere** instead of on the plane. It turned out that the **algorithm** for doing that, once the [Delaunay](https://github.com/Fil/d3-geo-voronoi) is computed, is just one line of code!","pinned":false,"hidden":false,"mode":"md"},{"id":49,"value":"$31","pinned":false,"hidden":false,"mode":"md"},{"id":57,"value":"$${link(4)}\n## *Day 4. Hexagons. “[Bi-Hexagonal Projection](https://observablehq.com/@visionscarto/bi-hexagonal-projection?collection=@visionscarto/30daymapchallenge)”*\n\nI’ve been fascinated by [Jacob Rus](/@jrus)’s art with high-precision **conformal** mappings, in particular his work on Schwarz–Christoffel transformations. Making this notebook was easy, in the sense that I just took Jacob’s *6gon* model and displayed it as a map **projection**—it literally took more time to write the text than to make the map work. You can play with the other models by replacing 6gon by any other of the values listed in [SCPie](https://observablehq.com/@jrus/scpie). Go ahead, I’ll wait here.","pinned":false,"hidden":false,"mode":"md"},{"id":64,"value":"$${link(5)}\n## *Day 5. OpenStreetMap. “[Paris, Texas](https://observablehq.com/@visionscarto/paris-texas?collection=@visionscarto/30daymapchallenge)”*\n\nI don’t often use OSM data directly, but I like OSMNx which gives me a clean **graph** representation of a city. Then I can play on it with **[Dijkstra](/@fil/dijsktra)**, to create **Voronoi**-like cells, etc. This is something I like to present in a workshop, because it doesn’t take a lot of code to go from “raw data” to a very nice experience (the wow effect). This notebook turned out to be popular too.","pinned":false,"hidden":false,"mode":"md"},{"id":70,"value":"$${link(6)}\n## *Day 6. Red. “[USSR, 1938](https://observablehq.com/@visionscarto/ussr?collection=@visionscarto/30daymapchallenge)”*\n\nA look inspired by a classic Atlas from the 1930’s. I usually don’t pay enough attention to the choice of colors; I enjoyed using these **warm tones**. The map is based an historical **shapefile**, and this was an opportunity to demonstrate how to read this format directly in Observable.","pinned":false,"hidden":false,"mode":"md"},{"id":86,"value":"$${link(7)}\n## *Day 7. Green. “[Hex NDVI](https://observablehq.com/@visionscarto/hex-ndvi?collection=@visionscarto/30daymapchallenge)”*\n\nThis could have been my entry for Day 4—Hexagons. It’s a mix between my previous [implementation in D3 of a three.js globe made by the BBC](https://observablehq.com/@fil/how-much-warmer-bbc) and a regular spherical point **distribution** based on [Gray-Fuller](https://observablehq.com/@fil/gray-fuller-grid). The notebook also uses the backface **projection**, and demonstrates how to read values encoded in a source image (so, it could have been an entry for Day 10—Raster).","pinned":false,"hidden":false,"mode":"md"},{"id":92,"value":"$${link(8)}\n## *Day 8. Blue. “[IGRF-90](https://observablehq.com/@visionscarto/igrf-90?collection=@visionscarto/30daymapchallenge)”*\n\nI didn’t particularly care for blue, so I made a map to demonstrate a spatial **interpolation** algorithm that I’m developing. The dataset is about the Earth’s magnetic field, and who can say if it is blue or purple? We have measured “true” values, which live on an irregular grid. We create a Delaunay **triangulation**, and apply barycentric interpolation inside each triangle. The method can be extended to extrapolate values (for points outside the convex hull of reference points), but I ended up publishing this partial implementation, in the hope that the simpler code is easier to read and understand than when it is mixed with extrapolation. However, the complete code is much more usable in practice—I just need to review and publish it when this challenge is over. If I forget and you’re interested, please comment here!","pinned":false,"hidden":false,"mode":"md"},{"id":100,"value":"$${link(9)}\n## *Day 9. Monochrome. “[Crater-shading](https://observablehq.com/@visionscarto/crater-shading?collection=@visionscarto/30daymapchallenge)”*\n\nTo be honest, this succession of color topics was becoming a bit boring. But there was a case to make for the Moon as being almost monochrome, and the elevation data as being greyscale. So I went for this demonstration of **hill-shading** on Luna. I tried to make a formula that was both understandable and fast enough (*i.e.* with not too many transcendental functions) to run at an interactive speed in the browser.","pinned":false,"hidden":false,"mode":"md"},{"id":113,"value":"$${link(10)}\n## *Day 10. Raster. “[Modified Stereographic Alaska GLSL](https://observablehq.com/@visionscarto/modified-stereographic-alaska-glsl?collection=@visionscarto/30daymapchallenge)”*\n\nThe [Modified Stereographic Alaska projection](https://observablehq.com/@d3/modified-stereographic-alaska) was missing from my [collection of GLSL](https://observablehq.com/collection/@fil/glsl-projections) **projections** — some kind of shifting with the original formula made the projection off by a few pixels. The challenge pushed me to “finish” this and hit the *Publish* button. I still don’t understand why my original code needed a correction—at least it’s shared now, and anyone can help.","pinned":false,"hidden":false,"mode":"md"},{"id":177,"value":"$${link(11)}\n## *Day 11. 3D. “[Volcano on A-Frame](https://observablehq.com/@visionscarto/volcano-on-a-frame?collection=@visionscarto/30daymapchallenge)”*\n\nI always struggle when it comes to 3D/VR. I have tried [a lot of things](https://observablehq.com/@fil/3d-apps-overview), but I never remember how [three.js](https://threejs.org/) does things and all the boilerplate. [A-frame](https://aframe.io) operates at an abstraction level that suits me better: the scene, objects, lighting &c. can be described (and animated!) as elements of the DOM, and A-frame then sets up three.js. Now I can focus on what matters (for me) — a series of data points in (x, y, z, color). Still striving for a tool that I’d really like.","pinned":false,"hidden":false,"mode":"md"},{"id":181,"value":"$${link(12)}\n## *Day 12. Population. “[Born in Antarctica](https://observablehq.com/@visionscarto/born-in-antarctica?collection=@visionscarto/30daymapchallenge)”*\n\nThis [tiny dataset](https://observablehq.com/@fil/nacimientos-en-la-antartida) of the 13 people ever born south of the Antarctic convergence is intriguing in itself—all the other continents have seen the births of hundreds of millions (or even billions) of humans. But it’s also fascinating with the *stories* that these births evoke. Who were these brave women—adventurous enough to go and give birth on one of the most remote and hostile places? What was the political and historical context which allowed militaries to weaponize these pregnancies for an elusive flag-planting goal? Most of them are still alive, and I hope that someone will write a book or film a documentary sometime to tell their stories.","pinned":false,"hidden":false,"mode":"md"},{"id":190,"value":"$${link(13)}\n## *Day 13. “[Natural Earth issues](https://observablehq.com/@visionscarto/natural-earth-issues?collection=@visionscarto/30daymapchallenge)”*\n\nI use Natural Earth **data** [all the time](https://github.com/visionscarto/world-atlas), but I’m somewhat frustrated by some of the choices and shortcomings of that project. I had this idea that **geocoding** all the open issues would give me a clearer view of disputed/discussed regions. It didn’t. The project failed, because my NLP skills are low, because my geocoding was poor, and because I wasn’t interested enough to spend more time on this notebook—I preferred instead to give a hand to Xavier Olive who was trying Observable for his [Day 14 entry](https://observablehq.com/@xoolive/disputed-territories)—which incidentally was about Natural Earth.","pinned":false,"hidden":false,"mode":"md"},{"id":202,"value":"$${link(14)}\n## *Day 14. New Tool. “[Hello, spl.js](https://observablehq.com/@visionscarto/hello-spl-js?collection=@visionscarto/30daymapchallenge)”*\n\nGeographic **data** comes in many **formats**: [GeoJSON](https://observablehq.com/@mbostock/geojson-viewer), [topojson](https://observablehq.com/@mkfreeman/topojson-simplifier), [shapefiles](https://observablehq.com/@visionscarto/ussr?collection=@visionscarto/30daymapchallenge), [geotiff](https://observablehq.com/@d3/geotiff-contours), [netCDF](https://observablehq.com/@fil/netcdf). We didn’t have the relatively new Geopackage (GPKG) format, but fortunately Jan Vaillant released *spl.js* just a few weeks ago, and now it works too! Yay! SpatiaLite!","pinned":false,"hidden":false,"mode":"md"},{"id":211,"value":"$${link(15)}\n## *Day 15. Map without a computer. “[Néguentropie](https://observablehq.com/@visionscarto/neguentropie?collection=@visionscarto/30daymapchallenge)”*\n\n*This project was not done on the occasion of the challenge.* We used a weird **projection** and **polygon clipping** to do actual wood cutting.\n\n

","pinned":false,"hidden":false,"mode":"md"},{"id":217,"value":"$${link(16)}\n## *Day 16. Urban/rural. “[Loir-et-Cher isodémographique](https://observablehq.com/@visionscarto/loir-et-cher-isodemographique?collection=@visionscarto/30daymapchallenge)”*\n\nI live in a very *rural* place, and all I see around is fields, cattle, forest, and not many people. But sometimes, when it comes to election maps, or jobs, or health… what the cartographer has to show is [people, not land](https://observablehq.com/@karimdouieb/try-to-impeach-this-challenge-accepted). This notebook used techniques from **optimal transport** to reach that goal—and I believe that it is the first time this framework has been applied to cartography. I was very happy with the outcome, as well as the colors and all—and this is one of the notebooks of this series that has enjoyed a bit of popularity. There is certainly much more to invent and discover with this approach (making circles is just one option among many, and the Loir-et-Cher is just one place). If you have ideas or questions, please contact me!","pinned":false,"hidden":false,"mode":"md"},{"id":227,"value":"$${link(17)}\n## *Day 17. Land. “[Theoretical Land Cover](https://observablehq.com/@visionscarto/theoretical-land-cover?collection=@visionscarto/30daymapchallenge)”*\n\n*“The built-up area and equivalent radius of settlements can be estimated with considerable accuracy from a knowledge of the number of inhabitants,”* writes **Waldo Tobler** in “Satellite Confirmation of Settlement Size Coefficients” (1969). If the formula given in that article holds, we can apply it backwards and make a map of land cover based only on the population sizes of major cities. It works, sort of. This notebook allowed me to make contact with one of the researchers doing current work on this topic, which was great.","pinned":false,"hidden":false,"mode":"md"},{"id":239,"value":"$${link(18)}\n## *Day 18. Water. “[Five Oceans](https://observablehq.com/@visionscarto/five-oceans-topojson?collection=@visionscarto/30daymapchallenge)”*\n\nWhen working on the “Born on Antarctica” notebook (Day 12), I had struggled to find a shapefile of the Southern convergence. So I tried to create the file which would have helped me, and go back in time. The project is in an unfinished state, because it appears that the limits between the oceans are not so easy to decide. (And I could not go back in time.)","pinned":false,"hidden":false,"mode":"md"},{"id":243,"value":"$${link(19)}\n## *Day 19. Island. “[Pantelleria Joy Plot](https://observablehq.com/@visionscarto/pantelleria-joy?collection=@visionscarto/30daymapchallenge)”*\n\nThis simple elevation profile came as a byproduct of the (quite ambitious) notebook I was working on for Day 21 (see below). Isn’t it nice what we can do with a few lines of code in [**Observable Plot**](https://observablehq.com/@observablehq/plot)?\n\n

","pinned":false,"hidden":false,"mode":"md"},{"id":251,"value":"$${link(20)}\n## *Day 20. Movement. “[Routing dollars](https://observablehq.com/@visionscarto/routing-dollars?collection=@visionscarto/30daymapchallenge)”*\n\nI’m reusing one of the favorite tools from my toolbox: **Dijkstra**’s algorithm, and apply it massively to another question raised by **Waldo Tobler**. The goal is to **summarize**, on a map, a large matrix of movements.\n\n

","pinned":false,"hidden":false,"mode":"md"},{"id":141,"value":"$${link(21)}\n## *Day 21. Elevation. “[Deriving Pantelleria](https://observablehq.com/@visionscarto/deriving-pantelleria?collection=@visionscarto/30daymapchallenge)”*\n\nIn this notebook I compute **gradients**, **hill-shading** (again), and introduce a **streaklines** implementation based on a paper by Whittaker (1977). Call to action: *A possible extension might be to think about [different ways to render the flows](https://observablehq.com/@esperanc/flow-field-variations)—for example to create hachures instead of streaklines: this could work by limiting the streaklines to a given maximum length. If you give it a try, let me know!*","pinned":false,"hidden":false,"mode":"md"},{"id":274,"value":"$32","pinned":false,"hidden":false,"mode":"md"},{"id":147,"value":"$33","pinned":false,"hidden":false,"mode":"md"},{"id":285,"value":"$${link(24)}\n## *Day 24. “[Historical shapefiles](https://observablehq.com/@visionscarto/historical-shapefiles?collection=@visionscarto/30daymapchallenge)”*\n\nI prepared this notebook very early in the month, as an obvious extension of the USSR notebook. It shows how to use Observable’s reactive dataflow and native handling of zip’ed FileAttachments, and how to decode shapefiles. All with very few lines of code. Unfortunately the source files are not very reliable.","pinned":false,"hidden":false,"mode":"md"},{"id":289,"value":"$${link(25)}\n## *Day 25. Interactive “[Earth-Nose-Direction Game](https://observablehq.com/@ambassadors/earth-nose-direction-game)”*\n\n\nThe END-game! Collaboration with [Fabian Iwand](https://observablehq.com/@mootari/) and [Radamés Ajna](https://observablehq.com/@radames). The initial idea was to use OpenCV, which Fabian had just found how to include in Observable, but for some reason the initial code kept on crashing, so we included a tensorflow model instead, and just added a few crazy colors and a few tricks of game design. This was supposed to be “as addictive as 2048”, but apparently we’re not game designers. Try it though, it’s fun to play!","pinned":false,"hidden":false,"mode":"md"},{"id":296,"value":"$${link(26)}\n## *Day 26. Choropleth “[Ann Arbor pycnophylactic](https://observablehq.com/@visionscarto/pycnophylactic?collection=@visionscarto/30daymapchallenge)”*\n\nBooooooring. Making a choropleth map is one of these tedious school exercises, and while it can be important to know how to make them, one has to admit that they’re a bit boring. But once again, we can call Waldo Tobler! The geographer invented a way to smooth choropleths, without losing sight of their spatial definition. This implementation follows (more or less) the original **algorithm**, extending it to work on a generic **graph** of randomly **distributed** points. Also, just spelling out p-y-c-n-o-p-h-y-l-a-c-t-i-c makes me giggle.","pinned":false,"hidden":false,"mode":"md"},{"id":304,"value":"$${link(27)}\n## *Day 27. Heatmap. “[Spherical Signed Distance](https://observablehq.com/@visionscarto/spherical-signed-distance?collection=@visionscarto/30daymapchallenge)”*\n\nWhen [Vladimir Agafonkin](/@mourner) implemented this small and fast **algorithm**, I knew I had to make it work on the **sphere**, just like d3-geo-voronoi was a spherical extension of his [Delaunator](https://github.com/mapbox/delaunator) —and of D3’s [d3-delaunay](https://github.com/d3/d3-delaunay). I don’t know if it can be useful for anything, but I was happy to have a complementary way of looking at the [distance to shore](https://observablehq.com/@fil/distance-to-shore) question. (Does this qualify for *heatmap*, I’m not sure…)","pinned":false,"hidden":false,"mode":"md"},{"id":324,"value":"$${link(28)}\n## *Day 28. Not flat. “[Walk on Spheres… on a Sphere](https://observablehq.com/@visionscarto/spherical-wos?collection=@visionscarto/30daymapchallenge)”*\n\nI had already played with the [WoS](https://observablehq.com/search?query=WoS) **algorithm** last year, and decided it was time to make a spherical version. The code itself is so short that it’s almost embarrassing I didn’t do it before. And the results look (imho) very nice!\n\n

","pinned":false,"hidden":false,"mode":"md"},{"id":312,"value":"$${link(29)}\n## *Day 29. *null.* “[Potential dollar flows](https://observablehq.com/@visionscarto/potential-dollar-flows?collection=@visionscarto/30daymapchallenge)”*\n\nUsing *potential flows* to make a map that summarizes complex **movements** was another great idea of (you guessed it…) **Waldo Tobler**. Étienne Côme had already done the animation of flows as [*winds*](https://www.comeetie.fr/galerie/wind/), but he was still computing the potential in R—this was an opportunity to bring yet another of W.T.’s tricks to Observable.\n\n

","pinned":false,"hidden":false,"mode":"md"},{"id":121,"value":"$34","pinned":false,"hidden":false,"mode":"md"},{"id":126,"value":"$35","pinned":false,"hidden":false,"mode":"md"},{"id":271,"value":"$36","pinned":false,"hidden":false,"mode":"md"},{"id":537,"value":"----\n*thumbnails and likes*","pinned":false,"hidden":false,"mode":"md"},{"id":671,"value":"collection = {\n yield await FileAttachment(\"collection@1.json\").json(); // cached version\n\n // see https://talk.observablehq.com/t/embed-a-collection/3504\n try {\n yield await fetchp(\n \"https://api.observablehq.com/collection/@visionscarto/30daymapchallenge\"\n )\n .then((d) => d.json())\n .then((d) => d.listings);\n } catch (e) {}\n}","pinned":true,"hidden":false,"mode":"ojs"},{"id":551,"value":"function link(day) {\n const d = collection[day - 1];\n\n return !d\n ? \"\"\n : html`

${\n d.likes > 2 ? `${d.likes} ♥ ` : \"\"\n }

`;\n}","pinned":true,"hidden":false,"mode":"ojs"},{"id":663,"value":"import { fetchp } from \"@tomlarkworthy/fetchp\"","pinned":true,"hidden":false,"mode":"ojs"}],"files":"$W37","stdlib":"1","titling":"auto"}