A geographic information approach to analyzing and visualizing presettlement land survey records for vegetation reconstruction

With a goal to improve the usefulness and diminish the limitations of Presettlement land survey records (PLSRs) in vegetation reconstruction that is important for understanding human impacts on the ecosystems, this dissertation explored three issues: to examine the geographic characteristics and data quality issues of PLSRs; to determine whether coarsely-resolved PLSRs are adequate to characterize the spatial patterns of individual tree species over large areas; and to develop methods to reconstruct the presettlement vegetation of large areas using spatially continuous representations.; The first issue was addressed using frameworks from GIScience. The geographic characteristics of space, theme, and time, and the data quality issues of lineage, positional accuracy, attribute accuracy, logical consistency, and completeness of the PLSRs were examined. The PLSR studies were also reviewed based on the geographic characteristic they analyzed.; The second issue was addressed using the public land surveys of Minnesota. Geostatistics was employed to reconstruct individual species distributions from different data resolutions. Statistical analysis showed significant correlations between the distributions from the coarse resolutions and the distribution from the finest resolution. Average prediction errors varied primarily with decreases in species abundances and were little influenced by species' degrees of spatial clustering.; The third issue was addressed using the private Holland Land Company (HLC) surveys of western New York. The quality of the data was first examined. Second, an automated approach was developed to reconstruct spatially continuous presettlement species distributions. Third, the reconstructed species distributions were analyzed to obtain the vegetation community distribution that was then compared with existing vegetation maps of western New York.; The results indicate that: (1) data quality issues vary with the PLSR study purpose and the spatial extent of interest; (2) visual comparison and statistical analyses both support that coarsely-resolved PLSR data are adequate to represent the spatial pattern of individual species over large areas; (3) the positional accuracy issue of the HLC surveys could significantly influence the use of the data in environmental modeling; and (4) the presettlement vegetation at both the species and the vegetation community levels are easier to interpret as spatially continuous representations than as discontinuous distributions of symbols.