Hacker News new | past | comments | ask | show | jobs | submit login
Show HN: Plot modern addresses on Earth 240M years ago (dinosaurpictures.org)
384 points by typpo on June 11, 2018 | hide | past | favorite | 59 comments

Hi HN, I built this by adapting GPlates (https://www.gplates.org), an academic project providing desktop software for geologists to investigate plate tectonic data. I'm amazed that geologists collected enough data to actually plot my home 750M years ago, so I thought you all would enjoy it too.

Even though plate tectonic models return precise results, you should consider the plots approximate (obviously we will never be able to prove correctness). In my tests I found that model results can vary significantly. I chose this particular model because it is widely cited and covers the greatest length of time.

The visualization itself is open source, though I have not yet cleaned up and pushed the plate tectonics integration (working on that right now): https://github.com/typpo/ancient-earth

I'm curious about the accuracy of the visualization. Specifically, at around 20M yrs, one can clearly see the mediterranean sea. The wikipedia page for the Zanclean Deluge[1] says it's theorized to have happened around 5M yrs ago.

Does this mean that the data from gplates doesn't support that theory? Or is that just an artifact of how things were put together for this project?

Nonetheless, cool project!

[1] https://en.wikipedia.org/wiki/Zanclean_flood

This is a good point, and I think we'd have to take it up with Chris Scotese, the geologist who produced the rasters I wrap around the globe[1]. There are some details in the PDF included in the download, which explains some coloration choices based on elevation (search for "paleocoastline")[2].

That said, I think there is artistic license taken, and the wet Mediterranean might be an oversight or even a statement by the artist (I've gotten emails from angry geologists because the globe does not advance one theory or another). Part of Scotese's work also predates some of the research on this topic.

It would certainly be more accurate to show a simple plate diagram rather than an artistic version of the Earth, and it would make my life easier too because then I could show continuous plate movement for each year. But there is a certain thrill in seeing ancient earth all globe-like, so I stand by my choice to use these rasters :)

[1] https://www.earthbyte.org/paleomap-paleoatlas-for-gplates/

[2] https://drive.google.com/open?id=1-q0WIa7ofISFHyBe4UxvN8DIPs...

would probably need to look into that theory further. That map at 20Mya is consistent with the fact that Florida was under water during that time. The Gulf Trough, another waterway, was running above the Florida platform around that time up to ~40Mya.

Nifty work.

> Even though precise results are returned by the plate tectonic models, the locations are still approximate [...]. In my tests I found that model results can vary significantly.

One challenge with science education content is to avoid nurturing rich ecologies of misconceptions... to the degree possible given resource constraints and conflicting objectives. So here are some random brainstormy ideas.

Add small inserts with different plate models? To reduce the implied certainty. Uncertainty is usually handled poorly in science education, with unfortunate results for education and society. Perhaps pixelate the globe?

The false colors represent elevation (EDIT: no, Scotese's hand-drawn sort-of-elevation maps), not biome. So "green map" may represent a long-term tan desert. Perhaps choose a less pretty but 'less likely to be misinterpreted as biomes' colorization? Maybe as an option? A biome map would be great to have... but at least as of some years ago, it didn't seem anyone was attempting unified paleobiome map models. It seemed more lack of incentive than lack of data, at least for some periods.

Sea level and coastline position varies at high frequency. 12k years ago, the shortest walk from Boston to the sea, was due south past Long Island. A color scheme that deemphasized the coast would allow coastline variation to be averaged/blurred.

Cloud coverage and patters vary greatly across time. Pretty clouds over Pangean desert made me smile. Perhaps explicitly call them "Bogus Clouds"? :) Science education content frequently shows bogus things. Rather than difficult reform, simply becoming more transparent/honest and comfortable with calling them bogus might be a relatively easy but significant cultural improvement.

The star field is missing long-term features, like a visible galactic disk. Having a higher-density star map might also usefully obscure that the set of bright stars is changing. Or perhaps just another "Bogus" label? They're really quite handy. :)

If anyone has time on their hands, it might be neat to attempt a more principled representation. Perhaps realistic appearance, averaged over a few My? Average cloud cover maps are less pretty than weather, but plausible long-term ones might be derived from relatively simple paleoclimate models. Similarly for biomes.

Paleoglobes are traditionally very misleading. As we move forward, it could be nice to improve on that.

Thanks for these thoughts.

My view is that there's a balance to be struck here. A pixelated globe with blurred coastlines and averaged features full of "bogus" disclaimers would not get the same reception as what I posted here. It would be less likely to be used as a teaching tool and less inspiring to students.

The very least I can do, though, is be clear about what this visualization is and isn't. I spent some time expanding creating a Credits modal yesterday (instead of an alert box) and will use that space to add more of this information and disclaimers.

Nod. Apropos bogosity, the underlying idea is this. Science education illustrations often have some aspects which are crafted with great care for accuracy, and other aspects that are artistic-license psychedelic bogosity. But students often lack the understanding to distinguish which are which. Leaving their grip on reality quite compromised. So if you can do anything to make it easier for them - yay.

Is... is the rest of the site with the dinosaurs your work as well?

Yeah, I definitely went down a bit of a rabbit hole of random dinosaur discovery!

It's a really wonderful site and horrible clicktarpit. I'm curious whether you have metadata about when the images were made (originally, I mean, in the case of reproductions of older images). It'd be nice to see the evolution of models from 19th century proto-godzillas to modern fluffy derposaurs.

Yes, I built this! But I can't take credit for the artwork.

This is a great effort, but the spot lighting on earth obscures more than the pretty effect is worth.

Thanks. I removed the spot lighting on Earth - accidentally left that code from when I had a specular map. Also increased ambient light on the night side based on feedback below.

Love the map and amazed to see how similar the world looks now to when the dinosaurs went extinct.

The title implies I can type in my address, but on mobile I can't find anywhere to do that.

Thanks - it was hidden by media query because I couldn't fit it. Just made a change to show the address input on mobile and hide other various controls.

Is there any way to see any of this without WebGL? Do you perhaps have some screenshots?

Wow, that is a fantastic knowledge resource! Something I have long hoped to find.


Is it supposed to do something when I enter an address?

Yeah, I think I hit my 150,000 daily quota on the Google Places API.

Love the map! Anyway to view a non-globe version?

You can download the rectangular maps from this page: https://www.earthbyte.org/paleomap-paleoatlas-for-gplates/

There are about 90 maps for different time periods, many more than displayed on the globe.

You may also be interested in https://paleobiodb.org/navigator/, which is an excellent web tool that is geared toward scientists, and therefore displays paleogeography more cleanly and diagram-like.

This is really cool! Thank you!!!

I felt so silly trying to zoom in and see satellite images of my house as it was 240M years ago.

I was hoping that if I zoomed in enough I'd get a 3d street view of a bunch of dinosaurs walking around. I was disappointed.

Could be possible if a 4 light year wide telescope is sent via a worm hole to one of the galaxies in the Virgo cluster (65m light years away) and live imagery of Earth (not via worm hole) is sent back via the worm hole to Earth.

Not needed. Once we can determine the starting composition of the universe, and have enough quantum computing power, we'll be able to simulate the past... and the future too. So you can see what your home looked like in the past and what it'll look like in the future.

Maybe, maybe not. We do not yet know that our Universe is a closed system

I can relate to this...

I would love to see what a potential future Earth would look like, and where I currently stand will be

https://www.youtube.com/watch?v=uLahVJNnoZ4 is pretty cool

As a child I was fascinated by the idea that the cliffs of Moher (https://www.cliffsofmoher.ie/about-the-cliffs/geology/) stood at the mouth of a massive ancient river but I've never found a picture of the earth in the geological past which showed such a river.

Nor does this model. Anyone know enough about geology to explain why? Is it because rivers are not represented at all?

Cool! But the model of earth appears to experience gimbal lock - use quaternions for rotations.

So the land that I stand on today, India, was once part of Africa and America's landmass. Crazy!

Yeah. Madagascar was closer to Kerala than Delhi. Blew my mind!

I love that you can rotate the globe around and there is a position where you can only see water.


That's almost still true if you turn a globe to show the pacific.

One thing: At 540 My ago, you misspell "Cambrian Explosion" as "Cabrian Explosion"

It was a long time ago, we don't exactly know what exploded.

Can we extrapolate this into the future?

Update: found this video: https://www.youtube.com/watch?v=uLahVJNnoZ4

Wow! This gave me a whole new perspective on the Uruguayan coast, it's so cool to know that what I am seeing latched off the Antarctic peninsula!!! Specially since it so recognisable. Just a nice piece of trivia, but awesome for me.

Is there data telling how deep the oceans and tall the mountains were during those eras?

Fascinating! I tried to see how the Alps form but it's quite a complex process

The difference between 260M and 240M in the Western US is really big. My hometown, Flagstaff, Arizona, goes from being near the ocean to being hundreds of miles away from the ocean, like it is now.

Some explanation of what's going on in NA: https://en.wikipedia.org/wiki/Laurentia

Great. Suggestion-

1. Can you add the number of years ago (eg. 100,000, 1M, 20M, 200M) next to the periods in the dropdown for those not familiar with it.

2. Can you simulate the changes automatically so it evolves?

The Americas look like the 'old world' in this model, to my surprise, except Mexico looks new. Asia could use better lighting on the marble.

Nice work! I feel like it can provide some perspective about things like nationalism, etc. Like a minor version of the 'overview effect'.

Very hard to steer on a cell phone (Android, Brave browser) - pinch zoom kept flying me into the Cosmos.

Like the idea though - will try later from desktop.

Xiaomi redmi 5 plus with brave browser as well works fine though..

Do you plan to extend it to show the future?

It's really great to play with the million years selector and the up/down arrow.

Really like it! Especially the short explanation of the age is quite educating

This is awesome, nice job!

Does that mean that an ocean that big could produce super hurricanes?

Be good to be able to turn off current country outlines

Thanks, I learned so much from this.

Pretty awesome. Thanks!


why show modern borders for 240M old earth?

That's the whole point of it. I don't understand your question.

Applications are open for YC Winter 2023

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact