Workshop: Spatial Models for Distance Sampling Data

In October 2015 at Duke University main campus, David L. Miller of the Centre for Research into Ecological and Environmental Modelling (CREEM) at the University of St Andrews and Jason Roberts of the Marine Geospatial Ecology Lab at Duke led a 4-day workshop on the spatial modeling of distance sampling data with R and ArcGIS. Thanks to all who attended! A special thanks to Allison Besch and Laura Lipps from the Duke Environmental Leadership program for hosting the workshop!

Attendees at the 2015 spatial modeling workshop

The workshop was a detailed dive into the theory and practice of the two-stage “density surface modelling” approach of Hedley and Buckland (2004), updated with methods from Miller et al (2013)–that is, first modelling detectability via a detection function, then using it to create a detection-adjusted spatial model using generalized additive models (GAMs). During practical sessions, we first used ArcGIS and MGET to prepare a pair of NOAA line transect surveys for analysis and linked them to remote sensing data. We then shifted to R to fit detection functions and GAMs. After comparing models and selecting the best, we predicted it across the study area using the remote sensing images to obtain a density and uncertainty surfaces for the species of interest.

All of the workshop materials may be downloaded from the workshop website.

Distance sampling animation

Animation illustrating the probability of making sightings according to how distant they are from the trackline