author = {William W. Hargrove and Forrest M. Hoffman},
  title = {Potential of Multivariate Quantitative Methods for Delineation and Visualization of Ecoregions},
  journal = environmanage,
  volume = 34,
  number = {Supplement 1},
  pages = {S39--S60},
  doi = {10.1007/s00267-003-1084-0},
  day = 1,
  month = apr,
  year = 2004,
  abstract = {Multivariate clustering based on fine spatial resolution maps of elevation, temperature, precipitation, soil characteristics, and solar inputs has been used at several specified levels of division to produce a spectrum of quantitative ecoregion maps for the conterminous United States. The coarse ecoregion divisions accurately capture intuitively-understood regional environmental differences, whereas the finer divisions highlight local condition gradients, ecotones, and clines. Such statistically generated ecoregions can be produced based on user-selected continuous variables, allowing customized regions to be delineated for any specific problem. By creating an objective ecoregion classification, the ecoregion concept is removed from the limitations of human subjectivity, making possible a new array of ecologically useful derivative products. A red-green-blue visualization based on principal components analysis of ecoregion centroids indicates with color the relative combination of environmental conditions found within each ecoregion. Multiple geographic areas can be classified into a single common set of quantitative ecoregions to provide a basis for comparison, or maps of a single area through time can be classified to portray climatic or environmental changes geographically in terms of current conditions. Quantified representativeness can characterize borders between ecoregions as gradual, sharp, or of changing character along their length. Similarity of any ecoregion to all other ecoregions can be quantified and displayed as a  representativeness  map. The representativeness of an existing spatial array of sample locations or study sites can be mapped relative to a set of quantitative ecoregions, suggesting locations for additional samples or sites. In addition, the shape of Hutchinsonian niches in environment space can be defined if a multivariate range map of species occurrence is available.}
  author = {Forrest M. Hoffman and Mariana Vertenstein and Hideyuki Kitabata and James B. White and Patrick Worley and John Drake and Matthew Cordery},
  title = {Adventures in Vectorizing the {C}ommunity {L}and {M}odel},
  booktitle = {Proceedings of the 2004 {C}ray {U}sers {G}roup ({CUG}) Conference},
  dates = {17--21 May 2004},
  location = {Knoxville, Tennessee},
  day = 17,
  month = may,
  year = 2004,
  abstract = {Described here are the extensive efforts of the authors to modify the Community Land Model for vectorization on the Earth Simulator in Japan and the Cray X1 at Oak Ridge National Laboratory. This paper follows experimental results presented at the Cray Users Group (CUG) Meeting in 2003 (White, 2003). Presented here are the technical details of the old and new internal data structures, the required code reorganization, and the resulting performance improvements. Additionally, performance and scaling of the final Community Land Model Version 3 (CLM3) on the IBM Power4, the Earth Simulator, and the Cray X1 are compared.}