Modeling historic landcover: An evaluation of two methodologies for producing baseline reference data

Quantification of the distribution and extent of landcover prior to human d evelopment can contribute to the management of natural areas. Often histori cal aerial photos postdate the desired time period. We developed two modeli ng approaches for reconstructing 1920s landcover prior to the first availab le 1943 aerial photographs We concluded that one method had an advantage, b ut realized that availability of historic aerial photos would limit its use in many areas. For this reason, we applied both methods to the U.S. Geolog ical Survey 7.5-minute Courtenay quadrangle on Merritt Island, Florida (USA ) and compared results. Method 1 used existing maps (1943 landcover, 1900s city/transportation, and 1913/1974 soils) and assumed that areas of natural land-cover in 1943 would also have been natural in 1920. Natural landcover polygons from the 1943 map were used directly and the ancillary maps were used to map urban polygons. Areas mapped as urban or transportation on the 1943 and 1900s maps remained urban, while areas mapped as urban in 1943 and not urban in the early 1900s were modeled using soil/vegetation associatio ns. The second method used urban polygons and overlaid them with soil infor mation; the soil/vegetation associations were then used to model all nonurb an landcover polygons. The detailed map from method 1 predicted all but two types, while the second method's coarse map predicted seven fewer landcove r types than were mapped in 1943. Either method (depending on modeling date and resources) can be useful for providing objective reference data to gui de management of natural areas.