Jump to content

 
Photo

Proving the benefits of generalization

- - - - -

  • Please log in to reply
7 replies to this topic

#1
Dennis McClendon

Dennis McClendon

    Hall of Fame

  • Validated Member
  • PipPipPipPipPipPipPip
  • 1,083 posts
  • Gender:Male
  • Location:Chicago
  • Interests:map design, large-scale maps of cities
  • United States

David Medeiros and I got into a discussion in the complex printing thread that I thought deserved to be a new topic:

 

David Medeiros, in various posts that I've collapsed below, said:

 

Too much detail, even when it is accurate, can create a map that is less effective than one that is generalized to suit its purpose.

*

Krygier and Wood talk about it a bit in Making Maps (Map Generalization and Classification). They discuss the idea that maps work by "strategically reducing detail and grouping phenomena". It's the reducing detail we're mostly about here.

*

It is, to me, a bedrock concept within cartography but one that most GIS map makers have a very hard time with, especially in science based GIS work. The assumptions in GIS are a) GIS and GIS data are inherently very accurate, and B) that simplifying the display of your data is somehow dishonest.

 

Maps intended for sharing and communication end up with a mess of raw data on them that clouds the actual message being communicated and in the end there is a potentially misleading message of accuracy that may get transmitted when in reality GIS data can often be much less accurate then manual mapping, just a lot more precise.

 

For me the use of generalization hinges on the concept of the maps communication goal. If transmitting raw data is the goal then obviously don't generalize, but that's a poor use of a map in most cases as only the analyst really needs to have all the data at hand for their work. Instead the map should be communicating your results or findings and for that it's probably better to simplify.

 

 

 

 

 

I think nearly everyone would agree on the kind of smoothing that deletes outliers for boundaries such as animal ranges or forest cover, because those make the map more understandable at a glance.

 

Where I'm having trouble is justifying simplifying something like a municipal boundary or national forest polygon from 500 vertices to 50 when 500 is overly complex for the scale.  In my gut, I absolutely believe it's worth the effort and makes the map better, but I fear that it's really an aesthetic preference based in old production methods.  Same with watercourses: is there really any psychographic evidence that readers have an easier time following my smooth, sinous uniformly 4 pt Missouri River instead of one that USGS compiled at 1:10,000 that shows every sandbar and inlet?

 

ITdxgAs.png


Dennis McClendon, Chicago CartoGraphics
chicagocarto.com

#2
Gretchen Peterson

Gretchen Peterson

    Master Contributor

  • Validated Member
  • PipPipPipPip
  • 238 posts
  • United States

In your graphic, the scale is such that I prefer more nodes, not less. However, at a smaller scale, it is better to have more generalization for the reason that more nodes = more perceived, yet unintentional, thickness at oxbows.



#3
David Medeiros

David Medeiros

    Hall of Fame

  • Validated Member
  • PipPipPipPipPipPipPip
  • 1,088 posts
  • Gender:Male
  • Location:Redwood City CA
  • Interests:Cartography, wood working, wooden boats, fishing, camping, overland travel, exploring.
  • United States

I think your example is a bit more complex than typical in that you are not only talking about a change in detail, but in mode as well (detailed polygon to simplified line).

 

And it may be that the scale is such that the change is not warranted (this may be what Gretchen is alluding to).

 

A good example for looking at what happens to detailed data at small scales and how aggregating and simplifying effect the reading of data could be something like BLM land holdings. The patterns are such that at small scale it can be difficult to clearly read the data unless the checkerboard pattern is generalized in some way.

 

I guess this brings up the question of what exactly are we talking about when we say "simplification"? Generalization, simplification, and abstraction are not the same things here (to me anyway).

 

There's removing detail (extra nodes) but preserving the overall shape and complexity of the feature shapes as well as the number of features, what I think of as simplification.

 

Removing complexity in the feature shapes (smoothing lines, reducing the number of bends in a river or boarder, aggregating smaller features into larger representations, or removing extra features etc.), what I'd call generalizing.

 

Abstraction could be a much more dramatic change in representation where the displayed shape may have very little of the original feature shape in it but suggests the feature through symbols. This can involve simplification and generalization but probably goes much further.

 

And then there is classification which can be thought of as a method for simplifying data, but probably doesn't fit directly into what we are talking about here.

 

We also can't forget that scale is not the only variable here, map purpose plays a significant role as well. So your simplified river line may not be suitable for a hydrology map at 1:50,000 but could be preferable for a hwy map at the same scale.


  • loximuthal and hasecbinusr like this

GIS Reference and Instruction Specialist, Stanford Geospatial Center.

 

www.mapbliss.com

 


#4
razornole

razornole

    Legendary Contributor

  • Validated Member
  • PipPipPipPipPip
  • 452 posts
  • Gender:Male
  • Location:Ozark Plateau, Arkansas
  • Interests:Photography, Cartography, Down-river canoeing, Backpacking, Cross country biking, Geomorphology, Ornithology, Ecology, Quaternary, and last but first; drinking beer on the beach.
  • United States

I'm in agreement with what David said.

 

Purpose is such an important variable, and I would also add your target audience (although that may be part of your purpose). 

 

I primarily make large scale recreational maps.  My target audience are people who want to use the map to bushwhack in the wilderness along with the folks who want to use the map for Sunday afternoon hike, bike, or paddle on a Sunday afternoon.  I would never generalize a trail, a stream, a shoreline, private property lines, or an isohypse.  I will simplify them (save private property).  I don't have a problem generalizing roads, city boundaries, and other things that aren't important to people that are in the backcountry.  I hate when I purchase a map and someone has generalized the trails.  They may show a little wiggle in the trail and that represents 10 switchbacks.  I would rather know that there are 10 switchbacks.  It will help me pinpoint where I am and what I can expect in my hike.

 

Another case where a generalized map did not suit my purpose.  I was paddling a 50 mile section of an unknown river in WI.  I had a nice generalized map of the river that had the mileage between access points, campsites, and major rapids.  The cartographer had done what you did with the river, removed sand bars and generalized bends on the river.  Before the trip I took that map and used satellite imagery and added all the islands/sandbars and redrew some of the bends.  That allowed me to use that map for navigation, it was crude but it worked.  I didn't have to guess if I was approaching a class I rapid or a class III.

 

kru


"Ah, to see the world with the eyes of the gods is geography--to know cities and tribes, mountains and rivers, earth and sea, this is our gift."
Strabo 22AD

#5
Kartograph

Kartograph

    Legendary Contributor

  • Validated Member
  • PipPipPipPipPip
  • 320 posts
  • Gender:Male
  • Location:Berlin, Deutschland
  • Germany

Hi!

 

The question at hand is actually at least twofold:

 

- Are there cognitive reasons for generalization?

- Are their material reasons for generalization?

 

I would answer both with a resounding "yes!".

 

Cognitive load is a well known concept in psychology. For several visual artifacts it has been proven that the more complex the visual situation is, the bigger the cognitive load. Problem solving is made easier when the cognitive load is reduced. For Map-like graphics it has been established, that the "amount of vector stuff" in the map directly influences the cognitive load as well as solving spatial tasks. This can actually be measured, as has been done by several researchers. Removing vertices is one way of doing it. As cartographers we definitely want to keep functional detail and reduce non-functional detail, ofc.

 

On the material side of things, a very quick example: vector maps get hugely unwieldy if left un-generalized. Take any rendering of OpenStreetMap data. After a certain zoom level, the vector version of a given raster tile would be way too large for all current pdf viewers, for example.

Furthermore as you reduce scale, you will keep vertices that are below any visibility threshold. Filtering them would be the most basic and most material kinds of generalization

 

There are other way more subtle arguments and effects. But regarding "proof", I think these two elements should be enough.



#6
Matthew Hampton

Matthew Hampton

    Hall of Fame

  • Moderator
  • PipPipPipPipPipPipPip
  • 1,314 posts
  • Gender:Male
  • Location:Portland, Oregon
  • Interests:Playing in the mountains and rivers.
  • United States

Well put Kartograph.  I think it was Bertin who inferred the role of the cartographer is to the reduce the mental cost of understanding a map, which is this cognitive load.  Paper maps are stuck at one scale (though there are some exceptions), so it's important to refine the scale/generalization-simplification relationship for the users needs.  

 

The scale-dependance of these variables can be intertwined to maximize the efficiency a user understands the map, so on a multi-scale digital map - you can provide a balanced amount of visual material that successively becomes more detailed at larger scales.

 

In Dennis's example above, both work - but the USGS 10k would be more appropriate at larger scales and at much smaller scales I think it would be even better to use a smoother 2pt line like below.

 

Attached File  ITdxgAs2.png   36.04KB   1 downloads


co-cartographic creator of boringmaps.com


#7
Dennis McClendon

Dennis McClendon

    Hall of Fame

  • Validated Member
  • PipPipPipPipPipPipPip
  • 1,083 posts
  • Gender:Male
  • Location:Chicago
  • Interests:map design, large-scale maps of cities
  • United States

Thanks, all.  Yes we should definitely distinguish among generalization, simplification, and abstraction.  As for my example, I see it was substantially exaggerated in the graphic I posted; the scale of the map I was making was 1:3,500,000 and that reach of river will be less than 3 cm on the page.  In this case, the rivers are pretty minor to the main purpose of the map (posted below for anyone who cares).

 

Of course, the purpose of the map is the most important starting point for these decisions.  But there's also the nature of the dataset. Especially for streams, I worry that working from large-scale data introduces a false precision into many maps.  There's no point in misleading a kayaker in 2014 into thinking he can navigate by counting sandbars on a map digitized from aerials flown in spring 2008.  Better in many instances, I think, to just generalize something like the Río Grande into a blue squiggle.

 

I would like to hear more about the cognitive load research, particularly anything that's focused on maps.  That's the crux of the argument I want to make to GIS folks about why such niceties (timesinks) as manual redrawing could ever be worthwhile.

 

 

Attached Files


Dennis McClendon, Chicago CartoGraphics
chicagocarto.com

#8
Charles Syrett

Charles Syrett

    Ultimate Contributor

  • Validated Member
  • PipPipPipPipPipPip
  • 537 posts
  • Canada

The "cognitive load" issue has also been called "eye noise", and it has, in fact, been researched for quite some time. Here is an example from 1970. (It appears to be a scan of a manuscript copy, but I remember seeing the published version in Cartographica back in the day.)

Charles Syrett
Map Graphics
http://www.mapgraphics.com


  • Bogdanovits likes this




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users

-->