Home

Welcome to the dsm wiki!

dsm is a package to model the spatial distribution of animals using data collected from a distance sampling (amongst others) survey. DSM stands for density surface model, though we often use such models to calculate abundance the name has stuck.

dsm is an R package and is designed to be used with Jeff Laake's mrds package or Dave Miller's Distance package.

Pages in the wiki are listed below, according to theme. Please contact me if you find any errors (including typos!).

Getting started

• Spatial models for distance sampling data: recent developments and future directions -- "introduction to DSMs" paper, accepted for publication in Methods in Ecology and Evolution
• Example analyses
• dsm documentation -- online version of the documentation shipped with the package
• Data format -- how to get your data into the correct format to use with dsm
• What's new in dsm version 2? -- talk given at CREEM, University of St Andrews, 2012
• Getting the latest version of dsm
• Problems? For advice about DSM analyses and survey methodology try the distance sampling e-mail list. If you think you've found a bug in dsm, you can report it on github.
• (in)frequently asked questions -- record of answers to questions I have recieved via e-mailthat may be of more general interest.

Modelling particular kinds of data

• Dealing with unusual transect shapes -- what to do when your transects turn corners etc, including one-sided transects.
• Accounting for correlation in DSMs -- talk to RUWPA, University of St Andrews, 29 July 2013
• Presence/absence models -- COMING SOON
• Using data from strip transect surveys -- COMING SOON

More technical information

• Estimating variance analytically for predictions in GAMs/DSMs -- overview of how uncertainty can be calculated for predicted abundance. Also includes explanation of the "variance propagation" method for including detection function uncertainty in a GAM.
• Using bam to fit models in parallel -- for big, "information rich" models computation can be speeded-up using parallel processing.

Known issues/problems

• ## Warning: matrix not positive definite messages
• Error in if (x0[i] == 0) 1 else x0[i] : missing value where TRUE/FALSE needed message when using dsm.var.prop or dsm.var.gam