Lands in the United States without detailed soil survey coverage, defined as the Soil Survey Geographic Database or SSURGO product, are considered "not complete" (NOTCOM). There are currently approximately 430 million acres of NOTCOM remaining in the United States, with ~309 million NOTCOM acres in Alaska. Thus, Alaska makes up more than 70% of remaining NOTCOM acreage in the U.S. and yet little soil data is available in unmapped areas in the USDA-NRCS NASIS database. In this work, I explore aspects of soil/ecological field data collection and sampling design in remote areas of Alaska dominated by permafrost-affected soils. The first part of this work consists of collaborative sampling of black spruce forests and associated permafrost-affected soils in the central Copper River basin, Alaska in conjunction with students and staff from Ahtna, Inc., the Alaskan Native Corporation whose shareholders have lived on and managed those lands for thousands of years, and who are now interested in utilizing soil survey and carbon data to target carbon sequestration initiatives. After sampling sites under burned and unburned areas, we assess trajectories of soil carbon, permafrost and vegetation change following fire to inform future management strategies. The second portion of this work examined the utility of implementing a Conditioned Latin Hypercube Sampling (cLHS) methodology in order to locate optimal sampling locations in large, previously unmapped areas of remote Alaska. The cLHS used in this study was constrained by an inclusive cost layer representing an aggregation of real-world costs associated with travel within the Yukon-Kuskokwim Delta study area located in south western Alaska. To accurately represent costs associates with travel throughout the study area, a single inclusive cost layer was created by aggregating travel cost information from multiple modes of transportation including travel by helicopter, float plane, river boat, and commercial flight to villages located within the survey area. Performance of the cost-constrained cLHS method was analyzed by assessing the model’s ability, constrained and unconstrained, to identify sampling locations that adequately represent a suite of environmental covariates and their feature space. Detailed analysis of cLHS model results for suggested sampling areas, with and without cost constraints, will aid in the creation of realistic sampling strategies that consider real-world scenarios and the optimal allocation of resources.
In partial fulfillment of the requirements for the M.S. degree in the Graduate Program in Land and Atmospheric Science