rick@hanauma (Richard Ottolini) (05/03/89)
Many commerical and non-commercial groups are working on this problem. Among the furthest along and most public, although not highest tech, is the United States Geologic Survey. Some issues: (1) The input problem: The USGS is committed to a decade long project to enter its topo maps into the computer. They are digitizing the originals--- about a dozen mylar layers per map each representing a feature: water, buildingds, contours, etc. Vector-extraction is semi-automatic; there are confusing features either due to complex geographies or original error requiring human intervention and expert system resolution. (2) The memory storage problem: I heard an estimate of ten terrabytes for for the 55K USGS map collection. And this is relatively coarse vector information. By the time the topo maps are digitized, ten terrabyte optical systems should be available. (3) The information representation problem: The USGS data is stored as vector sets in approximate location order. There is minimal feature naming or indexing. This fine for drawing maps, but difficult for locating named features or constructing roadmaps, etc. It is the classical database problem of efficiency verses functionality, but on a huge scale. (4) The copyright problem: When efficient GIS's become available, the databases themselves become exetremely valuable and some commercial mechanism will have to be found to distribute them. (5) The display problem: Maps require large, high resolution screen or printed displays. They need good interfaces, either interactive or well thought out artistry, to mange the complexity of information. I believe this problem is closest to a solution in both its traditional printed maps and computer interfaces. Some GISs are IMAGE-oriented or combined image and vector representations, derived from airplane and satellite images. They share, if not increase, the above problems.
littauer@uts.amdahl.com (Tom Littauer) (05/11/89)
In article <1997@Portia.Stanford.EDU> rick@hanauma (Richard Ottolini) writes: ... about the US Geological Survey and Geographic Info. Systems ... >(2) The memory storage problem: I heard an estimate of ten terrabytes for >for the 55K USGS map collection. And this is relatively coarse vector information. >By the time the topo maps are digitized, ten terrabyte optical systems should >be available. We used USGS 100K map data for our UniForum demo (you tell us your favorite obscure place, we'll produce a map) and can give you a little data... The USGS isn't used to talking about how much data is in the database in terms of bytes. Any estimates you get from them will be VERY rough. One data point: Transportation layer(s) at 100K took about 8-10 GB for the contiguous 48 states. As you get it from the USGS it is *NOT* efficiently encoded (ascii, fixed-length records, blank padding). The 8-10 GB I mentioned is as it came from USGS -- we didn't bother compressing because we were making a point about bigness :-) >(3) The information representation problem: The USGS data is stored as vector >sets in approximate location order. There is minimal feature naming or indexing. >This fine for drawing maps, but difficult for locating named features or >constructing roadmaps, etc. It is the classical database problem of >efficiency verses functionality, but on a huge scale. The companion Geographic Names Information System (GNIS) database serves as index (~180K town and locale names, MANY more features) but is accurate only to seconds of latitude/longitude. This gets to be a problem too... >(4) The copyright problem: When efficient GIS's become available, the >databases themselves become exetremely valuable and some commercial mechanism >will have to be found to distribute them. The 8-10 GB and GNIS data mentioned above ran about $40K >(5) The display problem: Maps require large, high resolution screen or printed >displays. They need good interfaces, either interactive or well thought out >artistry, to mange the complexity of information. I believe this problem is >closest to a solution in both its traditional printed maps and computer interfaces. The display problem drove me nuts - people not in the graphics business were impressed, but ... -- UUCP: littauer@amdahl.amdahl.com or: {sun,decwrl,hplabs,pyramid,ames,uunet}!amdahl!littauer DDD: (408) 737-5056 USPS: Amdahl Corp. M/S 278, 1250 E. Arques Av, Sunnyvale, CA 94086 I'll tell you when I'm giving you the party line. The rest of the time it's my very own ravings (accept no substitutes).