||SNIC Medium Compute|
||Tom Lindström <email@example.com>|
||2020-09-30 – 2021-10-01|
This project for Wildlife Management, Monitoring, and Modeling focuses on applied population and community modeling, specifically statistical modeling. We will develop a statespace modelling framework for population and community processes and parameterize models based on multiple data sources, including hunting data, traffic–wildlife accident and observational data from Svensk Fågeltaxering. Models will account for climate factors, necessitating inclusion of climate data and projection.
The project focuses primarily on Swedish ungulates, which include important game species as well as species that cause damage to agriculture and forestry. Forecasting these population and understanding how species interact are essential for a multitude of societal functions. Robust and reliable forecasts must therefore include all available data. Statistical modeling will be based on Hierarchical Bayesian inference, which typically is computationally demanding. At least in the early stages, we will perform computation with Stan, a software for Bayesian computation based on Hamiltonian Monte Carlo methods. Stan is efficient for shared memory parallelization. We will use RStudio and the package rstan as interface.