View Full Version : World building using ArcGIS

03-27-2008, 02:32 PM
Nothing better then using 20000 software for your roleplaying needs.
Sorry for the lack of pictures or screen shots at the moment, but i'm at school and don't have the project here

As i've stated in other threads, i'm in the process of creating a new realm.. start of which, i'm creating a world... Right now, i've come up with some basics for the mainland, and am planning to use ArcGIS to add the features in the world, so that they are geographically sound. I don't know how much everyone knows about Geomatics and analysis here in the guild, so i'll try to expain what I can...

First, to make most of my ideas work, i need to start with a DEM or Digital Elevation Model... what i've used for this is something people are firmiliar with here... Bryce... more admittedly I'm a novice at bryce, so i used photoshop to create a bump map, and added it into a bryce terrain... which i then exported as a USGS DEM. Playing around with the values so they made sense, i did things like added erosion, noise, smoothed etc... and the final product looks pretty good.

From there i imported it into ArcGIS as a raster. from here i can make contours and such, which i did, and found out that where the land meets the ocean (grey meets black) the countours were very square and ugly.. the contour above that looked quite good, so i preformed a calculation to lower the raster by 100.. (subtract 100 from the values)

after this, i had to make a polygon of the landmass (using the rasters contour) and made it a raster, then preformed multiply with the lowered raster (basically, the polygon raster had a value of 1 for land, 0 or everything else... and when multiplied with the lowered raster, it ends up with all the same values inside the polygon, and a zero everywhere else... (this is used to subtract everything below 0.

This is all i have done so far, but i have various plans for other things.

Rivers and watersheds
ESRI has a nice hydrology setup, which can allow me to find out what direction water will flow on every pixel in the raster... from that information we can then find where water would gather and run (rivers) on the entire raster... and when you give a point anywhere on the raster, you can extrapolate the watershed that point would fall under... this means i can create rivers on the world, using the DEM, instead of guess where they should be.

Climate is a bit more tricky, and has to start with placing points around the map with temperatures... A bit of guesswork here, and will be as detailed as you want to make it... but remember this is a fake place, so any "errors" don't have to be errors at all... from the points you can interpolate into a raster, and create a base climate for your entire map...

Plants and Animals
Plants can be found quite easily, using some simple analysis... basically, start with taking the slope and aspect of the DEM raster.. this gives you the steepness of the land, and the direction it faces... after that, its creativity... decide what your plants require to grow (some other facets and samples like soil types and such may need to be created, much like the temperature, but it is recommended to be more detailed)... with this information like for instance "cliff ferns require Xslope, and have to be south facing, on rocky soil and a temperate climate"... From the rasters we've created, we can preform an analysis to find areas that have all these attributes... giving you the areas this plan would thrive... from ther you can subtract any areas you wish (and add any areas) coming up with your final product... Same goes with animals, but you can also use the plants you create for their survival as well.

from that information, you can more easily create cities and regions, and populate your city in a human sense more completely. I believe using GIS software to figure out all this information is quite usefull, and creates a cohesive product, instead of finding conflicting ideas at a later date due to human error. The main thing to remember, is its fantasy... so nothing is an error, but being a geographer, i like my worlds to be as realistic as possible.

I hope this gives you an idea of what Geomatics and GIS can do for your roleplaying needs (and in the real world) any other ideas and suggestions would be much appreciated!

03-27-2008, 03:37 PM
So the preamble sounds good.....

Lets see some maps :)

03-27-2008, 04:20 PM
I had a VERY limited experience with GIS back when I worked for the Water Resources Board. They used it to not only map the entire state but also estimate water flowing into and out of a watershed. They even used it to simulate the flow of water throughout an aquifer based on the geology of the area.

As a Graphic Designer I mostly used it to get eye catching maps for the various publications. It's an extremely powerful tool and I am anxious to see your results.

03-27-2008, 04:42 PM
That's a very interesting use of ArcGIS! I like that you aren't guessing, but actually using physical geographical analyses to create the world. You could also use cost distance/ shortest path analyses between cities to create roads and paths. Cost distance could also be used to locate cities close to water and timber resources. And viewshed analyses to create watch towers with warning fires and/or burial or religious locations (for example, there was a study way back that found burial cairns to be related to the amount of open ocean visible and (if I remember correctly) visibility to each other, but I may be remembering incorrectly).

03-27-2008, 05:33 PM
You should put this in as a suggestion to Torq's Guild Map project!

03-27-2008, 05:37 PM
Hi Loogie,

This will be interesting to see what results you come up with. Many of the things you mention have been talked about (not so much on this site tho) especially the watersheds and the calculations of water flow.

I am part of a small group doing much the same thing but for Middle Earth. I remember one of the team mentioning ArcGIS - I think it might have been SeerBlue but I could be wrong. Anyway - the point of this project is to create a DEM of middle earth to a high degree of accuracy to the numerous reference maps and also try to conform to real geography as much as possible. You can take a look at the progress here - www.me-dem.org (http://www.me-dem.org)

I like what your saying with the ideas about calculating RPG stuff from physical properties. I was mentioning this a while back in another thread - here (http://www.cartographersguild.com/showpost.php?p=11871&postcount=6) you can check out the whole thread for a little more but these musings were a little bit of a distraction from the general map discussion.

Well, I am in anticipation of some nice pics :)

03-27-2008, 06:02 PM
Given the economic impact of a tool like ArcGIS, I'm impressed by the work, but might not find it altogether useful. Except for the, always fun, "gee whiz," factor. A good Bonehead Guide tutorial on the use of GRASS for things like this would be most appreciated, however :) .

Keep up the good work, though, and be sure to send pictures.

03-27-2008, 06:10 PM
Well here are some real quick ones... first, the export of the world from photoshop... theres some editing i needed to do to get things just right, but since this was already a jpg, i figured its easier then exporting in bryce.

the second is a screen from ArcScene, showing the continent in 3d, blue background just so you see everything better...

and finally, an screen from ArcMap, after i ran the flow direction and flow accumulation tool, you can see the rivers are coming along nicely. Lakes however will be a bit of a question... i'll have to use the flow accumulations to find out where water runs without going anywhere, and basically work out how much water and therefor how high the lake will be...

03-27-2008, 06:14 PM
lol i found grass earlier this month downloaded it and... well.. didn't touch a thing after that.. i have no idea how to use it, but if i got the basics down, and this type of analysis was available, i'd be happy to

03-28-2008, 01:39 PM
I was hoping someone who had a basic grounding in using at least one GIS program might be able to help me figure out how to use this thing. It seems promising as hell and you can't beat the price, but I can't even figure out how to load a raster.

The quick starts I've found seem to assume a basic grounding in GIS.

03-28-2008, 03:07 PM
well, if you point out the quick starts i'll take a look and see what i can figure out... i couldn't even turn it on when i looked at it (not that i tried very hard)

it does look promising (acctually better in some respects then a lot of other program i've seen) and of course.. free would make it an amazing addition to a roleplaying cartographers arsenal.

03-29-2008, 02:38 PM
Here is one good site (http://mpa.itc.it/markus/osg05/neteler_grass6_nutshell2005.pdf). I still can't get my own images imported, but at least I've been able to play with Spearfish...

03-31-2008, 01:55 PM
sorry man, i can't even get grass to run

no progress with arcgis as well.. i've been to busy with school..

working on a project that was due this morning at 8am...

oops... once everything slows down i should be ok tho...

03-31-2008, 03:32 PM
GRASS. I get excited every time I read about it, download it, and try to use it... too painful!

04-01-2008, 03:25 AM
I got frustrated and went back to playing with landserf. Not as capable, but at least I can import my #$@ing pngs and use them as HFs. My original intention was to use landserf to see if I could convert a png HF into something GRASS could see, but I couldn't face the ordeal and played some with landserf's feature extraction and such.

I'm really interested in these flowmaps, so I'm working on a q and d java implementation of a "flowmap." I'm not sure if it's technically a flowmap, it basically just sets the value of each pixel equal to the number of pixels that are directly above it(ie those from which water could flow into it.) I have three models: Greatest Drop('Water' only flows down to the lowest neighbor below a high point. Possibly analogous to SFD?), All Drops('Water' flows to all neighbors below a high point. One possible analog to MFD, perhaps?), Scaled Drop('Water' flows into all neighbors below a high point in a quantity proportional to the change in altitude. That is steeper paths will get more 'flow'. Another possible analog to MFD?) Applying a threshold to the resulting, "flow map," could be used to give a fairly decent idea about rivers. I assume this isn't really a flowmap in the sense GRASS uses the term, because the GRASS flowmap is deadly slow and this should be pretty quick. Initially I thought I'd have to implement a heap to get best performance, but it looks like it would be O(n log n) time whether I use a heap-based priority queue or just sort cell coordinates by height initially. I'm leaning toward the sort-and-go method, because my head just refuses to wrap itself around the bottom up method for building the heap.

It's a sign of how much fun I've had trying to get GRASS to work that I'd rather program my own solution :-)

04-01-2008, 12:15 PM
well arcgis is much simpler to use.. and i'm assuming most of whats in grass is the same if not copied off ESRI (they're the largest, and most used GIS software providers in the world... and i'm working for them! WOOT!)

though i don't know much about the math involved, i know the flow accumulation processes in arcgis have 2 main steps... you must create a flow direction grid(raster) which does what it sounds like you suggest takes every cell in the grid, and finds out which direction water would flow off it (N S E W + diagonals). that takes a short bit of time (i upped my anti in my map and made a new one thats 4096x4096) and in my example it takes about 2-3 mins... the result... isn't all that good for determining rivers (i guess it depends how you display it) but then you run the flow accumulation, which (from what i can disern) will go through the raster cell by cell and total those directions... so as one raster flows into another the next value is 2, the next is 3 etc etc... that takes a long time... i started running it about 10 mins ago now and its still going... last time i did it on a raster less then half the size i left and had some food, before coming back

those tasks don't seem terribly difficult to preform (generally most GIS is simple, its just that by hand it would take forever, by a computer, its fast) but the other thing that makes arcgis' tools so great is their options, for instance, flow accumulation can be weighted, meaning you can add acctual precipitation data to the formula, and get fairly accturate water fall results... as well as do other things (thats the only thing i could really think of)

the GIS tools will definitely help with many aspects, from finding trails, viewsheds (thanks for the ideas, never thought of em!) as well as doing climate modeling, and finding the best locations for various plants and animals.

now, if i only had the time to acctually do it :P

04-01-2008, 12:51 PM
You're not alone in preferring to program rather than try to learn GRASS. I do all my spatial modeling using a combination of Python/Numpy/GDAL/OGR/Shapely/SQLite. I liked Arc/Info and ArcView 3.2, but not ArcGIS. Even though I have a copy on a Windows computer in the lab, all I use it for is to look at maps and export them to Adobe Illustrator as soon as possible. I'll quit using it totally (except probably for teaching, can't find a way out of that yet) as soon as either Quantum GIS is further improved or I finish programming my Mac GIS visualizer.

04-01-2008, 01:35 PM
Well I did some more work on it... just playing around really, just seeing what i can do...

This is a sample hillshade from my new raster (much higher resolution)

After going over it visually, I preformed the same flow dir and accumulation tools

As you can see, the dir is almost the hillshade, making it much nicer to look at... at lower resolution its hard to see what this even does.

Heres the new flow accumulation.

The accumulation is more of a guildline.. all the lines shown here aren't suitable for rivers, they're mearly where water would collect (meaning drainage)... most of these lines would not even appear as a river in real life, just a path the water follows when it rains... being that these are fudged rasters and the analysis can't be 100% there are plenty of errors and such... but it gives you a good idea how to plan your rivers (and lakes)

Here you can see i've began building the rivers and lakes, i use the drainage layer, hillshade, and contours as guides, and do general outlines where i believe lakes and rivers should go... there are a lot of landlocked lakes here as well, but in real life there are a lot as well. This isn't an exact science, i just wanted it to make more sense then a fractal landmass, randomly generated with mountains and valleys exploding out of everywhere. So i've quickly traced some lakes and rivers (i may go back and do a better job, depending on how it turns out, maybe even just smooth these)

Anways, thats enough for now, i gotta get to school

04-02-2008, 11:52 PM
i've been using python a bit recently in class, and my coop will be pretty much all python as far as i know.. and i've begun to like it, its quite simple and the learning curve is through the roof...

i acctually like arcMap and arc gis, and its mostly because i've like ESRI as a company for a long time, they've been good to me in opportunity, support, and friendliness... but if your a mac boy, i understand the issues there...

the new software everyones talking about is called manifold... its MS friendly, but dirt cheap... (245 for a full pro license)... and apparently has a lot of features... so its cheap.. till you learn that its not very user friendly, and to get ANY support on it costs a lot of money... we're talking 30 per email, and 125 for a half hour phonecall... and thats per issue, so if you have multiple ones your paying for more... also, they are probably the cockiest company i've every seen.. if you look at their website its full of words like "the ultimate" and "worlds best"... they even say "it blows away the competition!" and "none of the other systems even come close!" and the funny bit is that its nothing special. they have nice looking pics of their maps, cause they got pros to make em.. and i recently did a project that required us to recommend a GIS service, and it required web servers. and well. if you look at the examples of web servers for manifold... they are terrible.. like simple java based screencaps that update based on what button you press...

thats enough ranting i guess :P

04-03-2008, 09:10 AM
...and the funny bit is that its nothing special.I couldn't comment a great deal - not having used it. But I do know that (im pretty sure right now) is the only full app in existence that is using the new nVidia CUDA interface as a math extension (not DX10 or physics) which can leverage a lot of extra power from your PC. MeDem have been hot and cool over CUDA for about half a year. It was 'cool' until last week until the GX2 card was released and its warming again... I don't have any involvement with them but I know what you mean about their personality... nothing short of obsessive MS fanbois :) (Probably undeservedly harsh there heh heh) But heck, they have achieved a feat in programming for CUDA so I give them that respect.

04-03-2008, 02:21 PM
I've been doing some more work on lakes and rivers as of late, and i hope to have some more screenshots to put up... then i'm going to start working on a climate, and possibly some soil models.

04-03-2008, 05:30 PM
There's a river draining the eastern lake that looks like it must be flowing uphill at one point between the saddle of two hills. Assuming precipitation exceeds evaporation, lakes should rise until they spill over the side.

Did you use the Bryce erosion tools on this? With the spine-like walls and slightly stairstepped channels it kind of looks like you did an iteration of erode, smoothed and then used erode again. Actually the spinewalls seem to come from the eroded tool. It's nice to see someone else using Bryce's erosion tools. On a PC, you have access to some really nice erosion tools, though. GeoControl is probably the best, followed by l3dt, but WorldMachine and Leveller are pretty good too. Probably better than the Bryce tools.

GRASS isn't working for me, so I've taken to playing with a java app called Landserf, and a UNIX toy called Terraform.

Landserf is excellent, I can import RGBA images as 32-bit elevations, and it has some excellent landform extraction tools. If nothing else I might be able to convert my 16-bit pngs into something GRASS can digest. Landserf doesn't really have any hydrology tools, though.

Terraform is a toy, pure and simple, but it's sea-level flattening tool is very useful. Better than the Wilbur exponential filter for creating continental shelves and beaches.

Erosion in terraform is problematic.

MFD flowmap doesn't work. Period. SFD does, but... it's SFD.

The river tool would be better if I could just create a pure river mask with it.

I'm playing with diffusion mediated by a flowmap with... interesting results. On a large scale map, it would make a nice swampland effect. On larger maps it could be masked out to create Chesapeake Bay type areas. Interesting.

What I'd like is an app that I could left click a point on the map and it would shade in the area where a river could conceivably flow from there. Then I could right click inside the shaded area and it would reduce the upstream area to paths a river could reasonably take to get from the first point to this point and then shade the potential downstream area.

04-03-2008, 07:06 PM
thanks for the ideas..
your assumtions on bryce were spot on.. and i had lots of trouble getting it to look right.. mainly, the erode isn't exactly what i'd like to see, mainly that it differs so much by resolution... (highest resultion made crazy spline ridges that were to numerous and unrealistic, and to low was just a big large square erosion almost completely ignoring the map i had already... so i found my self switching back and forth constantly

i'm definitly going to look at some of the programs you mentioned, i'd like to try to find some much better apps for this type of thing

04-03-2008, 08:58 PM
Terraform is a toy, pure and simple, but it's sea-level flattening tool is very useful. Better than the Wilbur exponential filter for creating continental shelves and beaches.

Have you tried the Remap Altitudes filter (Filter>>Other>>Remap Altitudes...) in Wilbur? It will let you apply any profile you care to draw to the surface.

04-05-2008, 01:17 AM
Next time I have access to the Cow Computer at work, I'll try that out. I'm really not ragging on Wilbur. I love Wilbur. I'm considering investing in an emulator just to run Wilbur. Well that and, you know, WM, GC, Leveller, VD, GTS, L3DT... um, et cetera. Perhaps what I should have said was, "even better."

Actually, I kind of overlooked Remap Altitudes 'cause I thought it as a terracing filter and I've never really been impressed by terracing except in certain specialized applications. The terraces always seem a little too uniform. Now that you mention it, though, that's exactly right for shoreline flattening. One thing I was thinking about was to take the original HF add in a fairly low-frequency moderate amplitude multifractal, terrace that and then subtract the multifractal back out. That might even be good for the shorelines, might produce interesting cliffs and the like. I need to make a note to do that on my next Moo Machine run.

Again, I'm not ragging on your program. I apologise if it appeared that way.

04-05-2008, 02:36 AM
Again, I'm not ragging on your program. I apologise if it appeared that way.

I wasn't annoyed or offended in any way. There are many peculiar features hidden away in the corners of that program and most of it is at best poorly documented and I wanted to be sure that you realized that there is a tool that might be easier to use than you had suggested.