![\"\"](\"https://me.darribas.org/uploads/2026/f03f3bcdb3.png\")
This is one of the clearest and shortest ways I’ve seen to reflect how I feel about using AI in work
\n\n\nGood writing is a tool to clearly communicate ideas from your brain into someone elseās.
\n
\n\nItās allowed to use AI tools while brainstorming or drafting your writing and certainly while proofreading, but make sure to consider the following principles while doing so:
\n
\n\nYou must stand behind every idea and every sentence in your docs.
\n
\n\nWriting is thinking.
\n
\n\nMore time should be spent authoring a document than consuming it.
\n
\n\n", "date_published": "2026-08-17T11:46:10+01:00", "url": "https://me.darribas.org/2026/08/17/there-are-no-lossless-transformations.html", "tags": ["Links"] }, { "id": "http://darribas.micro.blog/2026/08/03/summer-corner-pyrenees-august.html", "content_html": "Longer is not better.
\n
Summer corner.
\nPyrenees, Augustā26.
\n![\"\"](\"https://me.darribas.org/uploads/2026/30bfcd7ab1.jpg\")
\n",
"date_published": "2026-08-03T18:21:04+01:00",
"url": "https://me.darribas.org/2026/08/03/summer-corner-pyrenees-august.html",
"tags": ["Photos"]
},
{
"id": "http://darribas.micro.blog/2026/07/28/ive-been-out-on-leave.html",
"content_html": "I’ve been out “on leave” (holiday!) while this #GLaDpodcast dropped but, in case you didn’t listen to it, you absolutely should. Sarah is one of the most inspiring figures at the intersection of cities, tech, and society. Come for a peak of what it means to be Sarah Williams!
\n\n", "date_published": "2026-07-28T16:53:37+01:00", "url": "https://me.darribas.org/2026/07/28/ive-been-out-on-leave.html", "tags": ["Notes","Academic"] }, { "id": "http://darribas.micro.blog/2026/07/19/summer-rain-fog-maine.html", "content_html": "Summer rain fog.
\nMaineā26
\n![\"\"](\"https://me.darribas.org/uploads/2026/86dc45d812.jpg\")
\n",
"date_published": "2026-07-19T12:36:16+01:00",
"url": "https://me.darribas.org/2026/07/19/summer-rain-fog-maine.html",
"tags": ["Photos"]
},
{
"id": "http://darribas.micro.blog/2026/06/10/alien-data-in-urban-science.html",
"title": "Alien Data in Urban Science",
"content_html": "Iām on the train back to Liverpool after spending a couple of days in Bristol at the launch of the new Planetary Urban Science Lab (+Lab). It was extremely stimulating, if a little intimidating, to be among such a cadre of urban minds. As part of this, we were tasked to provide one slide and present it in three minutes capturing one āunsolved challengeā in Urban Science. This was good fun to prepare and I thought Iād write here about mine.
\n\n\nMost of the data cities need will be alien
\n
My challenge statement was: āmost of the data cities need will be alienā. Letās unpack it a bit. Testing theories, managing systems, and planning interventions in cities has always required data. This is not new and, in fact, it has often been the main blocker for progress. Not revolutionary, but worth stating explicitly.
\nThe good news is that we live in the most datafied age for cities and society. Thereās more, more frequent and, arguably, better data about cities and the activities that take place within them than ever. More importantly, every indication we have about the near future is that this trend will continue.
\nThe challenge in front of us is that most of these data are not like the traditional data we are used to when working on cities. Historically, most urban data tend to be tabular: spreadsheets about neighborhood characteristics, individual surveys, cadastral records. Street networks and associated graph representations are a shiny exception. The vast majority of data being generated about cities today are not tables: satellite images, text corpora (e.g., reviews from app data), street-level photographs, sound snippets. I argued these data are alien. I picked āalienā because it felt catchy and memorable -āunstructuredā might have been more accurate, but do you remember the last talk you saw about unstructured data? Me neither. I also think the term captures well what I wanted to get across, which is how different and impenetrable these data feel.
\nThis āaliennessā comes in, at least, two ways. One is because urban scientists, by and large, are not trained or used to working with these types of data. Images, for example, are very different animals from tables and spreadsheets. If all you know is the latter, the former feels too far from your comfort zone to do anything. The second source of āaliennessā stems from the fact that, more and more, the way we interact, process, and work with these data is mediated through AI (which, some people, refer to as an āalien intelligenceā). One of the things thatās become clear over the last 10-15 years is that the most useful way to work with these data is through neural networks. This used to be by turning them into specific summaries (e.g., number of cars in each photo, geometries of building footprints in a satellite image) but, more and more, it is through general-purpose, numeric representations (e.g., embeddings). These are built by the neural network to compress signal into a small footprint, and they are becoming extraordinarily good at it. But nothing in this process prepares them for human consumption. An embedding provides, e.g., 64 numbers that capture a great amount of statistical information about the pixels in an image. But each of those numbers on its own has no meaning for a human. It does not capture the number of cars in the image, or the amount of green space.
\nHow we learn to work with data that we do not fully understand but know is relevant for our purposes is the million-dollar question. I donāt know how we solve this, but I do know this challenge is going to become more and more relevant as more of the data we need take this shape.
\nUPDATE - The videos of all the lightning talks (and a snazzy one of the event) are now available here.
\n", "summary": "My one slide at the +Lab launch in Bristol.", "date_published": "2026-06-10T14:07:34+01:00", "url": "https://me.darribas.org/2026/06/10/alien-data-in-urban-science.html", "tags": ["Posts","Academic"] }, { "id": "http://darribas.micro.blog/2026/05/27/new-gladpodcast-episode-embeddings-for.html", "content_html": "New #GLaDpodcast episode: embeddings for the rest of us! Join us for a (hopefully) “lay enough” conversation about one of the most interesting ideas in geo-data that the last few years have produced.
\n\n", "date_published": "2026-05-27T11:40:28+01:00", "url": "https://me.darribas.org/2026/05/27/new-gladpodcast-episode-embeddings-for.html", "tags": ["Notes","Academic"] }, { "id": "http://darribas.micro.blog/2026/05/22/no-filter-like-the-southern.html", "content_html": "No filter like the Southern Europe filterā¦
\nLāAquila, Mayā26
\n![\"\"](\"https://me.darribas.org/uploads/2026/d0d6f6e82b.jpg\")
\n",
"date_published": "2026-05-22T12:26:28+01:00",
"url": "https://me.darribas.org/2026/05/22/no-filter-like-the-southern.html",
"tags": ["Photos"]
},
{
"id": "http://darribas.micro.blog/2026/05/08/happy-release-day-to-those.html",
"content_html": "Happy release day (+1) to those who celebrate! Third week in a row shipping imagery-based data products. One is a fluke, two is a pattern, three is real team effort. Imago releases CLiVE Air Temperature and Precipitation:
\n\n", "date_published": "2026-05-08T11:30:22+01:00", "url": "https://me.darribas.org/2026/05/08/happy-release-day-to-those.html", "tags": ["Notes","Academic"] }, { "id": "http://darribas.micro.blog/2026/05/07/talking-about-babies-and-everything.html", "content_html": "Talking about babies (and everything around them) was trickier on tape than on paper, but I think it was worth the effort, audience be the judge! New #GLADpodcast on academic babies:
\n\n", "date_published": "2026-05-07T11:15:08+01:00", "url": "https://me.darribas.org/2026/05/07/talking-about-babies-and-everything.html", "tags": ["Notes","Academic"] }, { "id": "http://darribas.micro.blog/2026/05/05/what-battersea-power-station-taught.html", "title": "What Battersea Power Station taught me about the value of satellite embeddings", "content_html": "Sometimes, magic finds its way through the forest of math that underpins most of my research. I spent a good amount of time a couple of weeks ago working on our embeddings workshop for GISRUK. We just released a new data product with Google satellite embeddings for small areas in the UK, and we couldnāt wait to take it on the road. This post is a note about one of the most insightful moments I had while preparing the materials but, if you want to check out the entire workshop, itās all open at:
\n\n\nhttps://imago-sdruk.github.io/embeddings_workshop/jupyterlite/content/02-Change.html
\n
One of the most appealing characteristics of satellite embeddings is their ability to explore change. Since embeddings map the information in an image to a shared latent space, it doesnāt matter where or when that image was taken, it all gets encoded in the same mathematical ālanguageā. Exploring whether the location of an image is similar or not to that of a different image; or whether such location has changed compared to what it looked like at a different point in time, is relatively straightforward. You can play with the ācomparison across spaceā idea in our Imago UK embedding Explorer. And we played with change across time in this notebook.
\nThe part that got me is the last exercise. We ask the question what area has become the most like Hyde Park between 2020 and 2024? This seems a bit esoteric but, I think, is an interesting one to ask. We know areas change within a city all the time, wouldnāt it be cool to know which ones are becoming more like some specific landmarks? In the workshop, I called it āHyde Park-ificationā, but I may deny this in publicā¦ It also turns out this is relatively straightforward to check with embeddings. You can follow all the details, code included, in the notebook. The gist is you calculate how similar all areas are to Hyde Park in both years, take the difference of that distance, and pick your winner as the smallest of those.
\nHereās our winner1:
\n
For the uninitiated reader, this is Battersea Power Station. Which, even if all you know is the first photo in the Wikipedia page I just linked to, youād be forgiven to think this embeddings sorcery is nonsense. This is as far as you can get from Hyde Park in terms of look and feel. So, whereās the catch?
\nThe catch, of course, is that we are not looking into overall similarity, but change. Our approach looks for areas that, over the period we consider, have become more like Hyde Park. That does not mean they are like Hyde Park. Thereās another interesting bit we show in the notebook: Battersea Station, by the standards of the areas we use, has changed very little. Again, this is counterintuitive maybe, but not incompatible with what weāre asking of embeddings in this exercise: you can change very little, but entirely in one particular direction.
\nAs it turns out, this is what happened to our winner. This insight did not come painlessly to me, as I struggled through a good 20/30 minutes of scratching my head about whether I was asking too much of this technology (particularly when itās aggregated from the pixel to an irregular area). Iām glad I stuck with this and followed it to the bottom. And Iām glad Google keeps funding Google Earth for free, including its time travel feature. Below is a comparison of the area in question between 2020 and 20242:
\n![\"\"](\"https://me.darribas.org/uploads/2026/pasted-image-20260415101159.png\")
If you canāt spot the difference (I couldnāt for the first ten minutes), follow the red arrow (which wasnāt there when I was looking for twenty minutes). The area, broadly speaking, has not changed much (which checks out with our low score of change). It was, and still is, a retired power station turned epicenter of cool south of the river. Mostly chimneys, mostly concrete. But thereās a small part that has changed. A small patch by the river used to be concrete buildings but, in between 2020 and 2024, it was flattened. Instead, it is now a patch of grass where Londoners enjoy the three days a year where the sun shines with all its fury3. You can see some of the before and after photos in this article by the BBC.
\nI love this example because it captures very well why Iām so excited about embeddings and about making them accessible to more people through data products like our small area one. I knocked out that notebook in about an hour (if you exclude my Google Earth rabbit hole). In that period, I was able to leverage data from seven massive satellite sources (see Figure S2 in this paper for details), across two periods of time on a setup that runs on my iPadās browser. Just let that sink in. What questions can we tackle that we couldnāt before? More importantly, which ones will we actually consider now that the price of asking is so low? I think weāre about to find out, and I canāt wait.
\nHappy release day to those who celebrate!!! Today, you can access our derived product that packages Google embeddings for small areas in the UK. Embeddings are changing how we work with satellite data. With this product, we hope it’ll be even easier to access and get started to look at places. If you take them out for a spin, let us know!
\n\n![\"\"](\"https://me.darribas.org/uploads/2026/0fc2a7c153.png\")
\n",
"date_published": "2026-05-01T08:56:01+01:00",
"url": "https://me.darribas.org/2026/05/01/happy-release-day-to-those.html",
"tags": ["Notes","Academic"]
},
{
"id": "http://darribas.micro.blog/2026/04/27/early-starts.html",
"content_html": "Early starts
\n![\"\"](\"https://me.darribas.org/uploads/2026/b484cfc8ea.jpg\")
\n",
"date_published": "2026-04-27T07:22:50+01:00",
"url": "https://me.darribas.org/2026/04/27/early-starts.html",
"tags": ["Photos"]
},
{
"id": "http://darribas.micro.blog/2026/04/24/happy-data-release-day-to.html",
"content_html": "Happy data release day to those who celebrate!!! We’ve been working really hard (how hard can it be to count clouds? a lot, it turns out…) to bring you all your cloudy needs neatly packaged. If you play with SPF, we’d love to hear from you!
\n\n![\"\"](\"https://me.darribas.org/uploads/2026/image.png\")
\n",
"date_published": "2026-04-24T13:24:58+01:00",
"url": "https://me.darribas.org/2026/04/24/happy-data-release-day-to.html",
"tags": ["Notes","Academic"]
},
{
"id": "http://darribas.micro.blog/2026/04/23/the-people-do-not-yearn.html",
"title": "š THE PEOPLE DO NOT YEARN FOR AUTOMATION",
"content_html": "Amen.
\n\n\n", "date_published": "2026-04-23T22:14:54+01:00", "url": "https://me.darribas.org/2026/04/23/the-people-do-not-yearn.html", "tags": ["Links"] }, { "id": "http://darribas.micro.blog/2026/04/10/pasa-la-vida.html", "content_html": "[21:51] I’m just saying these things aren’t everything, that not everything about our lives can be measured and automated and optimized. It shouldn’t be. And so the tech industry is rushing forward to put AI everywhere at enormous cost: energy, emissions, manufacturing capacity, the ability to buy RAM, locked into the narrow framework of software brain without realizing they are also asking people to be fundamentally less human. And then they’re sitting around wondering why everyone hates them. I don’t think a couple of haircuts are going to fix it.
\n
Pasa la vidaā¦
\n![\"\"](\"https://me.darribas.org/uploads/2026/image-20260410-093032-da8f460f.jpg\")
Iām just wrapping up a week in Berkeley and San Francisco. I came for two reasons, both have been fantastic, and Iām definitely feeling now the post-high blues that good academic interaction tends to induce.
\nI spent the first few days at the Berkeley Institute for Data Science, where the good folks of PySAL organised the Spatial Data Science Summit. This was a meeting to discuss opportunities and overlaps across the spatial data science community, and beyond, including the wider Python ecosystem for scientific computing. These days, Iām much less active (and thatās a kind way of putting it) in the development of PySAL. But it was nevertheless a fantastic experience. PySAL 1.0 shipped in 2009 and I was there to see it. Since then, itās been a project and a community that has brought me much of what I am and who Iāve become. And, as the clichĆ© goes, many of the collaborators have become close friends. ClichĆ©s are clichĆ©s for a reason.
\nThen I hopped on the BART to San Franciscoās Tenderloin, where geographers were descending left and right to discuss all things Geography at the AAG. I was part of a session brilliantly put together by Elizabeth Delmelle and Geoff Boeing on āproblem-driven methodsā (as opposed to the seemingly more common āmethod-driven problem developmentā) in the context of cities. I presented on satellite embeddings. When I was preparing the slides, I felt a bit uncomfortable because, arguably, I was about to engage in precisely the type of behaviour the sessions sought to avoid: āreaching for this yearās shiny new toolā instead of addressing cities' ‘wicked problems’ā. Elizabeth and Geoff prompted us to include a slide at the beginning stating the ābig urban problemā we were tackling. Iām so glad they did because it really nudged me to spell out why I think embeddings are more than this yearās shiny tool. My framing revolved around two key arguments. First, there is much more we can (and should!) do to tailor the data we use to the problems we tackle. Second, there is a lot of untapped data to help in that tailoring. In this context, satellites are one of those underutilised sources that can help provide better empirical fit to the questions we care about. And embeddings lower the barrier to access satellite data, making it cheaper to ask questions and explore ideas. I think, in the end, it went well and was well received.
\nI also participated in a panel organised as part of the discussion we had started in Berkeley earlier in the week around open source in spatial data science research. Serge Rey prompted us to think about existing gaps, low-hanging fruit, and surfaces of overlap. We covered quite a bit of territory, and Serge structured the conversation so there was clear and constant interaction with the audience, which became an āadditional panelistā. A lot of fun.
\nBesides the strictly āworkā things, this week also had plenty of space for fun. We (i.e., Rachel, Levi and your truly) attended the AAG Awards ceremony to pick up our Media Achievement award for GLaD. To celebrate it, that afternoon we hijacked a boardroom, stuck a hand-written note in the door that read ārecording in processā, and taped the first episode in a long while where we were physically together. And, after all that flurry of activity, the day after, I managed to convince Geoff Boeing and Martin Fleischmann to join me on a walk around very tall trees north of the Golden Gate. Long walks are underappreciated ways to exchange ideas.
\nWith that, the week came to an end. Itās been so much compressed in such a small amount of time and space. I know itāll take me a few days to unpack, literally and figuratively. This is conferencing at its best: more brain cycles in less days, all away than your usual routine. Academic life the way it should be.
\n", "summary": "Spatial Data Science Summit and AAG in seven days.", "date_published": "2026-03-23T13:01:00+01:00", "url": "https://me.darribas.org/2026/03/23/notes-from-a-week-in.html", "tags": ["Posts","Academic"] }, { "id": "http://darribas.micro.blog/2026/03/22/were-recruiting-for-two-phd.html", "content_html": "Weāre recruiting for two PhD positions at Liverpool with the Imago team and based at the GDSL:
\n\n\nThis project investigates how satellite-derived indicators can be co-produced with policymakers and communities to better reflect real needs and lived realities. Bridging Earth observation, participation, and policy, it aims to make satellite data more inclusive, credible, and impactful in decision-making.
\n
\n\nSpace2Health explores how satellite imagery can be transformed into meaningful measures of environmental exposures linked to health. The project relies on recent AI technologies to develop and evaluate satellite-derived indicators, examining how data choices and analytical approaches shape health-relevant evidence for research and decision-making.
\n
Both open until April 13th, get in touch for any queries.
\n", "date_published": "2026-03-22T21:16:45+01:00", "url": "https://me.darribas.org/2026/03/22/were-recruiting-for-two-phd.html", "tags": ["Notes","Academic"] }, { "id": "http://darribas.micro.blog/2026/02/16/this-happened-the-glad-podcast.html", "title": "The GLaD Podcast recognised with AAG award", "content_html": "This happened, the GLaD podcast is now the āaward-winningā GLaD podcast:
\nhttps://www.aag.org/2026-aag-awards-recognition/
Reproduced here for posterity:
\n\n\nAAG Media Achievement Award
\nThe 2026 Media Achievement Award is presented to Drs. Daniel Arribas-Bel, Rachel Franklin and Levi Wolf, the co-creators and hosts of the Geography, Life + Data (GLaD) Podcast. This podcast is celebrated for enhancing the understanding of geography by exploring the intersection of our discipline with data science, public life, and academiaāor, as their episode intro says, āgeography, life, geography life, and data. Launched in 2023, the GLaD Podcast and its predecessor series have produced over 50 episodes, amassing over 8,000 downloads, over 15,000 views on YouTube, and attracting more than 5,000 listeners worldwide. The podcast is renowned for its ability to simplify complex topicsāsuch as spatial data science and urban analyticsāthrough an engaging and accessible conversational style. It effectively breaks down barriers for students, early-career researchers, and non-specialists. Recognized as an invaluable educational resource, it has been integrated into graduate seminars and serves as a platform to humanize leading scholars. The podcast offers candid, practical advice on academic challenges like job searching and conference navigation, fostering a supportive community. GLaDās continued independent production underscores the creatorsā commitment to bridging the gap between academic research and the wider public.
\n
Like most things worth your time in life, we did not set out to do a podcast (the āIām in a bandā of the XXIst Centuryā¦) for the awards, but it does feel very good to be recognised, mostly by people listening to it every month, and now with this too.
\nThank you, thank you, thank you to everyone whoās clicked on the Play button at some point. And Thank you, Thank you, Thank you to the kind souls who nominated us for the award. Like Iāve said elsewhere, whatever they tell you, what academics really crave is peer recognition, and this feels pretty close to it!
\n", "date_published": "2026-02-16T13:04:56+01:00", "url": "https://me.darribas.org/2026/02/16/this-happened-the-glad-podcast.html", "tags": ["Notes","Academic"] }, { "id": "http://darribas.micro.blog/2026/02/13/new-episode-of-the-one.html", "content_html": "New episode of the one and only #GLaDpodcast. For this one, we dust off our oracles to speculate what 2026 (whatever is left of it anyway) has in store for Geography, Life, Geography Life, and Data! Come for the hot takes, stay for Polymarket-style futures contracts!
\n\n", "date_published": "2026-02-13T17:39:02+01:00", "url": "https://me.darribas.org/2026/02/13/new-episode-of-the-one.html", "tags": ["Notes","Academic"] }, { "id": "http://darribas.micro.blog/2026/02/10/this-was-a-lot-more.html", "content_html": "This was a lot more fascinating than I thought it would. Come for the craze around OpenClaw, stay for the background story of the mind behind it.
\n\n\nThen it will send you a message on WhatsApp, and suddenly you talk on WhatsApp, making this flow easy. That was hard. Yeah, even coming up with the idea that youāre not editing the configuration, because the agent can edit its own configuration. You don’t have to update anything because the agent can update itself. You can literally ask your bot to update itself, and it will fetch itself and update itself, coming back with new features. Planning the technical giveaway, so far that’s the magic.
\n
\n\nāI would recommend them to be infinitely curious. Yes, it’s going to be harder to enter this market. It’s absolutely going to be harder, and you need to build things to gain experience. I don’t think you need to write a lot of code, but you need to explore. Thereās a lot of open source that is complex, that you can check out and learn from.
\n
\n\nI added a bootstrap file to explain the model that is now being born to create an identity and a soul where the values of the user are, and the model will be like, “Hello! Who are you?” “Who am I? What’s my name?” You know, and I’ve watched people do it, and that’s where the magic starts.
\n
\n\nYou don’t have to update anything because the agent can update itself. You can literally ask your bot to update itself, and it will fetch itself and update itself, coming back with new features. Planning the technical giveaway, so far that’s the magic. That’s
\n
\n\nYou don’t have to update anything because the agent can update itself. You can literally ask your bot to update itself, and it will fetch itself and update itself, coming back with new features. Planning the technical giveaway, so far that’s the magic.
\n
\n\nYou don’t have to update anything because the agent can update itself. You can literally ask your bot to update itself, and it will fetch itself and update itself, coming back with new features. Planning the technical giveaway, so far that’s the magic.
\n
\n\nBut this new world needs people that can bridge both areas.
\n
\n\n", "date_published": "2026-02-10T23:32:18+01:00", "url": "https://me.darribas.org/2026/02/10/this-was-a-lot-more.html", "tags": ["Links"] }, { "id": "http://darribas.micro.blog/2026/02/04/how-markdown-took-over-the.html", "title": "š How Markdown took over the world", "content_html": "Even, you know, even when I get a pull request, I’m actually more interested in the prompts than in the code. I ask people to please add the prompts, and some do, and I read the Prompts more than I read the code, because to me this is a way higher signal of like how did you get to the solution? What did you actually ask?
\n
Sobre very good Internet History lesson, full of delicious details that corroborate the idea the web is built more on the nice part of human beings than the not so nice.
\n\n\n", "date_published": "2026-02-04T23:42:56+01:00", "url": "https://me.darribas.org/2026/02/04/how-markdown-took-over-the.html", "tags": ["Links"] }, { "id": "http://darribas.micro.blog/2026/01/28/yesterday-we-the-imago-team.html", "content_html": "But itās important for everyone to know that the Internet, and the tech industry, donāt run without the generosity and genius of regular people. It is not just billion-dollar checks and Silicon Valley boardrooms that enable creativity over years, decades, or generations āĀ itās often a guy with a day job who just gives a damn about doing something right, sweating the details and assuming that if he cares enough about what he makes then others will too.
\n
Yesterday, we (the Imago team) gave a workshop on satellite embeddings for a room with about 50 people from different corners of the UK Government. It was fantastic in every respect (with the possible exception of Github bringing down our website because it thought it was receiving a DDoS attack lol). It’s so exciting to see these ideas and technology move at the speed of light from state of the art research to Government. Anything we can do at Imago to facilitate that transfer, we’re here for that!
\nIf you’re curious, materials are available at:
\nimago-sdruk.github.io/EMBED2Soc…
\nAnd if this looks interesting to your and/or your organisation, do get in touch!
\n", "date_published": "2026-01-28T12:26:17+01:00", "url": "https://me.darribas.org/2026/01/28/yesterday-we-the-imago-team.html", "tags": ["Notes","Academic"] }, { "id": "http://darribas.micro.blog/2026/01/23/earth-embeddings-towards-aicentric-representations.html", "title": "š Earth Embeddings: Towards AI-centric Representations of our Planet", "content_html": "Very very timely paper that captures the current zeitgeist in EO and AI. If nothing else, it serves as a fantastic introduction to one of the technologies that I think(/hope) will help the most bring imagery to the masses in the coming years.
\n\n\nEarth embedding vectors emb are produced by a family of embedding functions E that map continuous location inputs (i.e., longitude, latitude with optionally elevation, and time) into a d-dimensional vector space:
\n
\n\nFigure 2: Earth embeddings provide different functions: (1) They compress high-dimensional data into a lower-dimensional vector format. (2) They fuse together different geospatial data modalities, from different types of images to text and tabular data. (3) They can interpolate to unseen spatiotemporal locations, where raw data is missing. (4) They are interoperable with other AI foundation models, such as LLMs, through aligned embedding spaces.
\n
\n\nas explicit models, extracting embeddings from raw data (e.g. satellite imagery) associated with a location (emb ā¼ Eexplicit(datalocation))
\n
\n\nimplicit models, returning embeddings from only location inputs (emb ā¼ Eimplicit(location)).
\n
\n\nEarth embeddings map places and times that share similar properties closer together in embedding space.
\n
\n\nGeoFMs are large-scale modeling and learning frameworks, whereas Earth embeddings constitute the interoperable, location-indexed data outputs that can be stored, shared, or queried independently of the model that created them.
\n
\n\nWe posit that Earth embeddings will emerge as the dominant format of geospatial data in the AI age
\n
\n\nways in which users can employ Earth embeddings for prediction, conditioning, simulation, and search
\n
\n\nCall to action: Advancing analyses and applications with Earth embeddings.
\nā¢ Evaluating and benchmarking Earth embeddings
\nā¢ Explainable and interpretable Earth embeddings
\nā¢ Learning planetary processes with Earth embeddings
\n
\n\nEarth Embedding Models: Explicit Feature Extraction versus Implicit Neural Representation
\n
\n\nChallenges and opportunities for improving Earth embeddings.
\nā¢ Model capacity
\nā¢ Spatio-temporal heterogeneity
\nā¢ Data curation and scaling
\nā¢ Learning objective
\n
\n\n", "date_published": "2026-01-23T22:35:26+01:00", "url": "https://me.darribas.org/2026/01/23/earth-embeddings-towards-aicentric-representations.html", "tags": ["Links","Academic"] }, { "id": "http://darribas.micro.blog/2026/01/21/a-bit-late-to-post.html", "content_html": "The research agenda we outline is fundamentally interdisciplinary: Earth embeddings will rely on feedback from domain scientists, e.g. in ecological, geological, oceanographic, and atmospheric sciences, that incorporate Earth embeddings into their analyses and from data practitioners apply- ing Earth embeddings in their workflows and products.
\n
A bit late to post about (though not to publish!), we have a new episode of the #GLaDpodcast out. If nothing else, be enticed by the title (the oldest profession in Geography?!?!?!); if something else, delight in Anthony Robinson’s views on maps, AI, and microwave ovens!
\n\n", "date_published": "2026-01-21T18:29:07+01:00", "url": "https://me.darribas.org/2026/01/21/a-bit-late-to-post.html", "tags": ["Notes","Academic"] }, { "id": "http://darribas.micro.blog/2026/01/19/from-data-dump-to-data.html", "title": "š§ From Data Dump to Data Product", "content_html": "So many common points and arguments that really resonate here and make me more hopeful for Imago. The discussion of data as infrastructure, invisibility as success, and thinking really hard about how to make sure āitāās not only here now, but tomorrow and the day after are points thatāll stay with me. And itās also great to find more people whoāre thinking creatively (not only from the tech side of things) to ensure the world has more collective-ness around data. Most recommended listen.
\n\n\n", "date_published": "2026-01-19T00:07:31+01:00", "url": "https://me.darribas.org/2026/01/19/from-data-dump-to-data.html", "tags": ["Links","Academic"] } ] }The way I frame it is like, the game is figuring out how to lower the cost of asking questions.
\n