missing from this environ- 

 mental puzzle. 



Researchers needed an 

 up-to-date land-use and 

 land-cover inventory, new 

 maps and a land classifica- 

 tion system to fill out their 

 resource data base. The 

 most cost-effective and 

 practical way to get them, 

 they decided, was with 

 Landsat digital data. 



Inventories had been 

 done before, Copeland 

 says, but Khorram and his 

 team had the right ma- 

 chines and expertise to 

 make this project work. 



Khorram 's team 

 collected data for the study 

 from five scenes by 

 Landsat, taking care to 

 match seasons, times and 

 tides. Then, pixel by pixel, 

 they studied how much 

 light had bounced from the 

 surface to the satellite and 

 found patterns in the 

 piecework. These patterns 

 told them what kinds of 

 vegetation, and in some 

 cases land uses, were present in the 

 Albemarle-Pamlico watershed. 

 Buildings and barren farm fields 

 reflected huge amounts of light, for 

 instance. Deep, clear water absorbed 

 light like a sponge. 



From there, they developed 20 

 new categories for land use and land 

 cover. The sharp eye of the satellite 

 now could distinguish between pine 

 forests and bottomland hardwoods, 

 for example. And it could differenti- 

 ate a low pocosin of wet, brushy land 

 from a low marsh of cordgrasses and 

 bulrushes. And Landsat could tell 

 them something else. 



The digitized spectral, or light- 

 wave, data could be used by map- 

 ping experts to draw detailed land- 

 use maps. 



Siamak Khorram 



The phofo-lihe images 

 satellites create can 

 give scientists a clearer 

 picture of how riverine, 

 oceanic and estuarine 

 systems intertwine 

 Researchers can ujffp 

 the data to countjp 

 wetlands at the coast 

 and map water palitq. 



If resource managers 

 want to protect lowland 

 forests, Copeland says, 

 mapping via satellite can 

 pinpoint these areas. Or, it 

 can designate highways, 

 farmlands, urban areas, 

 wetlands and other land 

 uses — pixel by pixel — in 

 a context that managers 

 may need to know. 



The researchers verify 

 their work by sampling 

 their findings in the field. 

 They match aerial photo- 

 graphs and county maps to 

 the computer's images to 

 identify plots of Atlantic 

 white cedar and pocosins in 

 Dare County, marshland in 

 Hyde County and riverine 

 swamps near Washington. 



Eighty to 90 percent of 

 the time the researchers are 

 right. The standard is 85 

 percent accuracy. 



"This is not a perfect 

 technology," Brockhaus 

 states. "You'll never get 

 100 percent accuracy. If we 

 get 85 percent, we're very 

 happy, very happy." 



Other than the sampling, the 

 researchers do not need to visit the 

 site. "That's the beauty of this 

 technology," Copeland says. 



Instead of taking water samples 

 from one point and then another, or 

 driving to one forest to check the tree 

 types and then another 50 miles 

 away, satellite data can fill in the 

 spaces between, giving a more 

 realistic picture. 



"If you can see it all, you can 

 begin to build that big picture," 

 Copeland says. "Otherwise it's very 

 difficult. If you've got to go down 

 here and look at the edge of the 

 Albemarle, then get in your car and 

 drive up to look at the edge of the 

 Chesapeake Bay, getting the big 



10 NOVEMBER/DECEMBER J 992 



