Joint forecasting of salmon lice and treatment interventions in aquaculture operations

Abstract

The need for joint forecasting of parasitic lice and associated preventative treatments stems from large monetary losses and reduced fish welfare associated with such treatments, and the distribution of potential future treatments can be used in operational planning to hedge their associated risk. We present a spatio-temporal forecasting model that accounts for the joint dynamics between lice and treatments where spatial interaction between sites is derived from hydrodynamic transportation patterns. The model-derived forecasting distributions exhibit large heterogeneity between sites at significant levels of exposure, which suggests the forecasting model can provide great value in assisting operational risk management. For the data used in this study, within a 20-week forecast period, our results suggest up to a 40% lice-induced loss in the value of farmed salmon, with differences up to 20%pt. in the average exposure of each site.