Error propagation in an integrated spatially-explicit individual-based model

Authors: KORALEWSKI, TOMASZ - TEXAS A&M UNIVERSITY; WANG, HSIAO-HSUAN - TEXAS A&M UNIVERSITY; GRANT, WILLIAM - TEXAS A&M UNIVERSITY; BREWER, MICHAEL - TEXAS A&M UNIVERSITY; Elliott, Norman - Norm

Submitted to: Ecological Modelling
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 11/5/2022
Publication Date: 11/14/2022
Citation: Koralewski, T.E., Wang, H., Grant, W.E., Brewer, M.J., Elliott, N.C. 2022. Error propagation in an integrated spatially-explicit individual-based model. Ecological Modelling. 475:Article 110215. https://doi.org/10.1016/j.ecolmodel.2022.110215.
DOI: https://doi.org/10.1016/j.ecolmodel.2022.110215

Interpretive Summary: Error propagation is an important consideration in individual-based modeling. We investigated the propagation of error due to uncertainty in the initial condition using a previously published spatially-explicit individual-based model that simulates sugarcane aphid infestations of sorghum. We used three neighboring cells as alternative initial infestation sites and analyzed the resulting model outputs in pairwise scenario comparisons. Resulting spatio-temporal patterns of aphid infestation, as estimated by timing and probability of first infestation, were different for the alternative initial infestation locations. The patterns were substantially different for two pairs of scenarios originating from neighboring cells, indicating that error propagation through the studied system depends not only on the extent of the initial spatial uncertainty, such as the distance between the model initialization sites in the pair of compared scenarios, but also on the actual location of the model initialization sites, likely reflecting differences in the characteristics of those locations. The spatio-temporal trajectories of the propagating error indicate that the system limits the error in the observed variables. The results of this study are important in that they demonstrate that the model functions sufficiently well that its use in assessing the ecological processes involved in sugarcane aphid population dynamics will lead to valid ecological insight.

Technical Abstract: Error propagation is an important consideration in individual-based modeling, and it has been considered insufficiently studied. We investigated the propagation of error due to uncertainty in the initial condition using a previously published spatially-explicit individual-based model that simulates sugarcane aphid infestations of sorghum. We used three neighboring cells as alternative initial infestation sites and analyzed the resulting model outputs in pairwise scenario comparisons. Resulting spatio-temporal patterns of aphid infestation, as estimated by timing and probability of first infestation, were different for the alternative initial infestation locations. The patterns were substantially different for two pairs of scenarios originating from neighboring cells, indicating that error propagation through the studied system depends not only on the extent of the initial spatial uncertainty, i.e., the distance between the model initialization sites in the pair of compared scenarios, but also on the actual location of the model initialization sites, likely reflecting differences in the characteristics of those locations. The spatio-temporal trajectories of the propagating error indicate that the system limits the error in the observed variables.