Improving Accuracy of Large-Scale Prediction of Forest Disease Incidence Through Bayesian Data Reconciliation
Date of Award
Master of Science (MS)
Mathematics and Statistics
Mevin B. Hooten
Mevin B. Hooten
Increasing the accuracy of predictions made from ecological data typically involves replacing or replicating the data, but the cost of updating large-scale data sets can be prohibitive. Focusing resources on a small sample of locations from a large, less accurate data set can result in more reliable observations, though on a smaller scale. We present an approach for increasing the accuracy of predictions made from a large-scale eco logical data set through reconciliation with a small, highly accurate data set within a Bayesian hierarchical modeling framework. This approach is illustrated through a study of incidence of eastern spruce dwarf mistletoe (A rceuthobium pusilfom) in Minnesota black spruce (Picea mariana). A Minnesota Department of Natural Resources (DNR) operational inventory of black spruce stands in Northern Minnesota found mistletoe ill 11% of surveyed stands, while a small specific pest survey found mistletoe in 56% of the surveyed stands. Through use of Bayesian data reconciliation, cross-validation shows an increase in agreement of the DNR forest inventory with the more accurate specific pest survey from 53% to 76%. Using this model, we predict 35% to 59% of black spruce stands in Northern Minnesota are infested with dwarf mistletoe.
Hanks, Ephraim M., "Improving Accuracy of Large-Scale Prediction of Forest Disease Incidence Through Bayesian Data Reconciliation" (2010). All Graduate Plan B and other Reports. 1235.
Copyright for this work is retained by the student. If you have any questions regarding the inclusion of this work in the Digital Commons, please email us at .