Soil Survey Disaggregation in a Low-Relief Depressional Wetland Landscape.

Efforts to utilize conventional soil maps in wetland conservation and restoration planning are often hampered by the coarse scale of SSURGO soil maps relative to the scale of restoration decisions, the spatial aggregation of soil components, and the difficulty in accounting for uncertainty in soil maps. To improve identification of hydric and other components on depressional wetland landscapes in the Mid-Atlantic Coastal Plain, we have developed a method to digitally disaggregate conventional soil map units into individual soil classes based on natural soil drainage class and particle-size family class. We are using training data collected in the field and environmental covariates derived from LiDAR, SSURGO, and other spatial datasets to build an ensemble tree classification model to predict soil classes. Model predictions will be validated using independent field data on soil morphological characteristics collected across a 543 km² watershed.