Outerra Worlds Forums

Please login or register.

Login with username, password and session length
Advanced search  

News:

SMF - Just Installed!

Author Topic: DEM for climate model  (Read 10405 times)

Dan

  • Newbie
  • *
  • Posts: 3
    • View Profile
DEM for climate model
« on: October 24, 2013, 08:14:12 pm »

Hi all,
I am a climate scientist at the University of Bristol, UK, and a huge Tolkien fan since an early age!  For some outreach activities, I am interested in running a climate simulation of Tolkien's Middle Earth.  My 'normal' work involves setting up the climate model for many past time periods of the Earth when the continents were in different positions, so doing the same for Middle Earth / Arda is not a huge leap!  In order to do this, I need a DEM.  In particular, a DEM (topography and bathymetry) at a resolution of about 720 x 360 points covering the 'globe' (any resolution is actually OK, as I can regrid).  Most formats would be OK (e.g. zipped ascii/binary).  I could probably work with a native ArcGIS format as well.  I would obviously acknowledge the project appropriately if it was successful.
Looking forward to hearing from you - hope this is possible!
Thanks a lot,
Dan
« Last Edit: October 24, 2013, 08:44:44 pm by Dan »
Logged

Redrobes

  • Administrator
  • Newbie
  • *****
  • Posts: 43
    • View Profile
Re: DEM for climate model
« Reply #1 on: November 01, 2013, 12:37:41 am »

Hi Dan, I run the build process for the final stages of the map and terrain generation. For this we use a simulation which I wrote called GeoTerSys which sorta stood for Geomorphological Terrain System. What it does it basically calculates the temperature, rainfall, evaporation and so on for the terrain and works out how that might affect the river flow and the vegetation for the land. Then we export those parameters into greyscale maps which are then fed into the programmable texture generator so that it can render out the final maps. Outerra doesnt use many of these but it is now using the vegetation map. Outerra has recently posted that its now doing some simple biome information so in the near future we will be using more of those maps to import information into Outerra's biome system to render the vegetation as well as land colour and so on.

We use a map of 40K x 40K pixels. Each pixel knows about the liquid content (rain, river, snow) and the temperature, vegeation as well as the height. Plus it has a number of other parameters all tied up in that pixel. We run a very simplified terrain and climate model since we have limited CPU power. At the mo it takes about 15 mins per tile and we have 1600 tiles so it takes a long time. The app is written so that it could in theory be done in parallel tho. But even so we have to manage the overlap of the tiles and ensure that data leaks off of one tile into the next adjacent one. So its not a simple process.

We could provide some kind of DEM for it. Monks usually handles that aspect of it and I think we can generate some kind of lower res version which is not so bandwidth heavy but right now he is a bit tied up with some other work. We do have above and below sea level heights for the map but you must understand that our map is but one interpretation of JRRT's map. We have tried to ensure that it is hydrologically correct tho even if the map is not one which is super accurate to any physical model when he sketched it.

Now I am not sure what a res of 720 x 360 points means. That seems so completely ridiculously low res to do any meaningful work with.  The map we use is 4000km square so your res would mean something like 5km per pixel. We originally aimed for 25m per pixel but were at 100m at the mo. Even 1km per pixel is quite low res tho.

Anyway - I have rambled enough.... your thoughts ?
Logged

Dan

  • Newbie
  • *
  • Posts: 3
    • View Profile
Re: DEM for climate model
« Reply #2 on: November 05, 2013, 02:44:11 am »

Hi,
Thanks very much for your reply - very useful!

The climate model I am using was developed at the UK Met Office.  It is essentially the same model that they use to produce weather forecasts, but running at a lower resolution.  It is computationally expensive (even at the low resolution I run at, it takes a week on a parallel supercomputer to simulate 50 years of weather - and several hundred years are needed to generate a meaningful climatology and to allow the model to 'spin-up' from the initial state of the atmosphere which is provided by the user).

Because the model is based on the fundamental equations of fluid mechanics and physics, the weather is all internally generated, and the only input file which the user supplies is the topography and bathymetry - everything else (temperature, precipitation, winds, clouds, humidity, pressure, vegetation....etc...etc) is predicted by the model. [actually, you also have to tell it some physical and geophysical constants like the acceleration due to gravity, molar gas constant, strength of the sun, amount of greenhouse gases in the atmosphere etc.]

So, yes, the resolution is low (and in fact the climate model runs at an even lower resolution than the 720x360 which is needed to generate the topography and bathymetry), but it is not dissimilar from other climate models which are run globally.

I wanted to start a test simulation, so I generated my own topography/bathymetry by tracing some of Fonstad's maps, and colour-coding for different heights, then converting to a bitmap and reading into some IDL code (which then adds some random noise, decreases bathymetry away from the coasts, and lowers rivers relative to the surrounding land).  The climate model is now running based on this topography, and is busy producing weather!  It is still early days, but the initial results look very interesting...I'll keep you posted!  It would be interesting to try a different input global map of heights/depths, so please do send me one if anyone has time to generate it!

Many thanks,
Dan



« Last Edit: November 05, 2013, 02:47:57 am by Dan »
Logged

monks

  • Administrator
  • Full Member
  • *****
  • Posts: 173
    • View Profile
    • ME-DEM Project
Re: DEM for climate model
« Reply #3 on: November 21, 2013, 12:42:34 am »

Hi Dan, sounds awesome that!  :o At some point hell yeh we'll send you a terrain map. I'm a bit busy for the next couple of months with another project I think but if you're still around we'd love to give it a whirl! Out terrain data is kinda locked down in a project on a backup drive- I've run out of space on my main drive so I can't load the project up until I've finished with this other project.

monks
Logged

Dan

  • Newbie
  • *
  • Posts: 3
    • View Profile
Re: DEM for climate model
« Reply #4 on: December 06, 2013, 01:00:26 pm »

Hi all,
So, we 'published' our results in the end, using  DEM I created by tracing some of the Fonstad and Tolkien maps.  See here:
http://www.bristol.ac.uk/news/2013/10013.html
Hope it is of interest...the results are presented in detail in the pdf available from the link above.
cheers,
Dan
Logged

Redrobes

  • Administrator
  • Newbie
  • *****
  • Posts: 43
    • View Profile
Re: DEM for climate model
« Reply #5 on: December 06, 2013, 02:46:54 pm »

Hi Dan,

Thats very cool. Sorry we were not able to provide some base DEM data for you but you seem to have generated some which provides some answers which make the world quite plausible. I dont think Tolkien himself worried about the realism to a great extent but its fun to see how it might play out in such a simulation.

From our point of view it may generate some items of interest that we could possibly incorporate into the MeDem map but mainly ours is generated from the simulation we run with a lot of user generated input such as the baseline temperature map. So if you have generated an objective model from no user input then that's quite a feat. That climate model must be quite complex.

Were still hoping to push out a new MeDem version of the map to the Outerra engine with its new base texture feature and programmable biome texturing thing it has now. So ill be sure to post about that in due course when Monks gets more free time from his commercial terrain thing he is tied up on.

I will have a more thorough look at the PDF but for now ill just say I always knew Texas was somewhere in Mordor....

:)

EDIT: Chaps at the www.cartographersguild.com would be real interested in that PDF. Its right up their street !
« Last Edit: December 06, 2013, 02:49:45 pm by Redrobes »
Logged