7 replies to this topic

[benbakelaar]

http://dsc.discovery...sprawl_tec.html

I'm interested in comments on the time period, 1976 to 1992, which in my opinion is much too short.

The following statement seems suspect:

"To produce meaningful data, the researchers merged high-altitude photography from around 1976 with satellite images from 1992 -- the most recent images available. They divided the space into 8.7 billion grid cells, each representing 900 square meters (a little less than a quarter of an acre)."

How much work is involved in using newer satellite imagery?

[Martin Gamache]

http://dsc.discovery...sprawl_tec.html

I'm interested in comments on the time period, 1976 to 1992, which in my opinion is much too short.

The following statement seems suspect:

"To produce meaningful data, the researchers merged high-altitude photography from around 1976 with satellite images from 1992 -- the most recent images available. They divided the space into 8.7 billion grid cells, each representing 900 square meters (a little less than a quarter of an acre)."

How much work is involved in using newer satellite imagery?

<{POST_SNAPBACK}>

Alot of work, but in this case I suspect they are using the 1992 NLCD which conveniently gives an urban class. There is obviously new imagery since then, but none mosaiced and classified for the entire country in one dataset, althought the circa 2000 dataset is almost complete now which will allow some comparison for 1992 - 2000 eventhough some of the classes have changed.

i suspect the earlier date is also dictated by the existence of a dataset or a complet photo set from that time period. You always work with the data you have especially in change detection work where you can't go back intime to create a dataset for a more interesting or convenient date.


mg

[benbakelaar]

martin thanks for that explanation. good to know these things.

[Matthew Hampton]

The USGS had a wonderful program that looked at urban sprawl (Urban Dynamics Research), that was unfortunately cut with the Republican administration. They made great headway that will (hopefully) get picked-up in 2008.

If you are interested I encourage you to take a look at their finished document. Their range of years is about the same (1970-1990).

Based on the first sentence in the Discovery article - they must have missed all the hard work Roger, Janis and William completed years earlier.

[Martin Gamache]

I meant to take a look at the original article to get a sense of what the original data was but did not get a chance to do that yet. As an aside...Geographer are usually suspicious of the methods used by economists when they enter the spatial realm without guidance from the geographers themselves...often for good reasons.

mg

[Matthew Hampton]

The USGS and Dept of Interior used GIRAS for the 70's and NLCD for the 90's.

It's is interesting to see the progress of the MRLC with respect to the 2000 era NLCD. I just checked their viewer and it is slowly coming along...

[Martin Gamache]

Here is a link to the original paper:
http://diegopuga.org/papers/sprawl.pdf

[Matthew Hampton]

Here is another link to some of his broader research.

He is purportedly doing research on landscape and obesity - it sounds very interesting.