NEFSC Seminar Series

**Speaker: Olav Breivik, Norwegian Computing Center**

**Title: "Time varying observation variance in stock assessment, and a spatial model for estimating abundance of marine mammals"**

**Abstract**

The uncertainty of log observations used as input to assessment models may differ between years. Currently, two procedures are implemented to accommodate for such time varying structures in SAM. One option is to inform SAM about estimates of the uncertainties, or we can assume a link between the expected observation and the corresponding variance. In this presentation, I will go through how SAM can accommodate for that the log observations may have different variances in time.

I will also go through a spatial point process model for estimating abundance of marine mammals. The model is developed in TMB and assumes that the marine mammals are Poisson distributed driven by a latent Gaussian random field. Line transect data is used, and we estimate both the detection function and the true underlying density simultaneously.