TY - JOUR
T1 - Development and assessment of regeneration imputation models for National Forests of Oregon and Washington
AU - Kralicek, Karin
AU - Sánchez Meador, Andrew J.
AU - Rathbun, Leah C.
N1 - Funding Information:
This research was supported by funding from Pacific Northwest Regional Office (Region 6) of the USDA Forest Service and by the ARCS Foundation Scholar program. We would like to particularly thank Tom DeMeo, Regional Ecologist of the Pacific Northwest Region, for his input on the grouping of plant associations into FPAGs. We would also like to thank two anonymous reviewers for their insightful comments. Northern Arizona University and Oregon State University are equal opportunity providers.
Publisher Copyright:
© 2017 Elsevier B.V.
PY - 2018/2/1
Y1 - 2018/2/1
N2 - Imputation models were developed to predict seedling regeneration density and composition on National Forest System (NFS) lands in Oregon and Washington. The models were based on Forest Inventory and Analysis and Pacific Northwest Regional NFS Monitoring data. Individual models were developed based on broad forest plant association groups (FPAGs) with all model development and analysis conducted in R using a most similar neighbor-like imputation approach. Model performance was evaluated based on bias, mean absolute deviation, root mean-squared error (RMSE), and error rate in correctly predicting the total presence or absence of any regenerating species regardless of species (Total ER). Low to moderate RMSE (≤7400 regeneration stems ha−1) and low to moderate Total ER (≤50%) were observed for 25 out of 58 FPAG-specific models. The regeneration imputation models produced in this study represent a large first step towards developing flexible, expandable, and adaptable regeneration models that can be easily incorporated into existing growth models like the Forest Vegetation Simulator.
AB - Imputation models were developed to predict seedling regeneration density and composition on National Forest System (NFS) lands in Oregon and Washington. The models were based on Forest Inventory and Analysis and Pacific Northwest Regional NFS Monitoring data. Individual models were developed based on broad forest plant association groups (FPAGs) with all model development and analysis conducted in R using a most similar neighbor-like imputation approach. Model performance was evaluated based on bias, mean absolute deviation, root mean-squared error (RMSE), and error rate in correctly predicting the total presence or absence of any regenerating species regardless of species (Total ER). Low to moderate RMSE (≤7400 regeneration stems ha−1) and low to moderate Total ER (≤50%) were observed for 25 out of 58 FPAG-specific models. The regeneration imputation models produced in this study represent a large first step towards developing flexible, expandable, and adaptable regeneration models that can be easily incorporated into existing growth models like the Forest Vegetation Simulator.
KW - Forest inventory and analysis
KW - Most similar neighbor
KW - Multi-species
KW - Pacific Northwest
KW - Plant associations
UR - http://www.scopus.com/inward/record.url?scp=85037540120&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85037540120&partnerID=8YFLogxK
U2 - 10.1016/j.foreco.2017.12.004
DO - 10.1016/j.foreco.2017.12.004
M3 - Article
AN - SCOPUS:85037540120
SN - 0378-1127
VL - 409
SP - 667
EP - 682
JO - Forest Ecology and Management
JF - Forest Ecology and Management
ER -