148-9 The Boundary Waters Canoe Area Wilderness (BWCAW) Digital Soil Mapping Project.

Poster Number 1203

See more from this Division: SSSA Division: Pedology
See more from this Session: Soil Survey Present and Future: II

Monday, November 16, 2015
Minneapolis Convention Center, Exhibit Hall BC

Suzann Kienast-Brown1, Tom D'Avello2, Danielle Evans3, Zamir Libohova4, Robert Long5, Jessica Philippe5, Roger Risley3, Larissa Schmitt3, Amy Koch6, Robert Vaughn7 and Joseph Brennan8, (1)USDA-NRCS, Bozeman, MT
(2)NSSC-Geospatial Research Unit, USDA-NRCS, Morgantown, WV
(3)USDA-NRCS, Duluth, MN
(4)National Soil Survey Center, USDA-NRCS, Lincoln, NE
(5)USDA-NRCS, Saint Johnsbury, VT
(6)USDA-NRCS, Hilo, HI
(7)USFS, Salt Lake City, UT
(8)USDA-NRCS, Saint Paul, MN
Poster Presentation
  • BWCA_soil_survey_SSSA_2015.pdf (5.0 MB)
  • Abstract:
    The Boundary Waters Canoe Area (BWCA) resides in the Superior National Forest in Northern Minnesota. The area is approximately 1.1 million acres, extending along the Minnesota and Canada border. The landscape is a complex mosaic of four soil parent materials; namely Rainy Lobe Till, lacustrine Agassiz sediments, outwash deposits and organic deposits. Soil drainage classes range from well-drained to ponded, and soil depth classes range from bedrock exposures to very deep deposits. Access to most of the project area is restricted to foot and paddle access only. A digital soil mapping project for the BWCA began in late 2012. Training data was collected during the 2013 field season. Several digital soil mapping approaches were employed and evaluated to create raster-based soil class maps including; unsupervised and supervised classification, knowledge-based classification and regression trees, logistic regression, random forests, and expert, heads-up digitizing. Eleven soil classes were mapped using these approaches. During the summer of 2014 validation points were collected in the field for an accuracy assessment and to instruct the possibilities for post-processing and delivery of the final raster product. Accuracies are currently being evaluated and preliminary results indicate an overall accuracy (and khat) that varies from 0.71 (0.67) to 0.53 (0.48) depending on the input parameters.

    See more from this Division: SSSA Division: Pedology
    See more from this Session: Soil Survey Present and Future: II

    << Previous Abstract | Next Abstract