Distance Sampling Simulations

Performs distance sampling simulations.It repeatedly generates instances of a user defined population within a given survey region, generates realisations of a survey design (currently these must be generated using Distance software in advance < http://distancesampling.org/>) and simulates the detection process. The data are then analysed so that the results can be compared for accuracy and precision across all replications. This will allow users to select survey designs which will give them the best accuracy and precision given their expectations about population distribution. Any uncertainty in population distribution or population parameters can be included by running the different survey designs for a number of different population descriptions. An example simulation can be found in the help file for make.simulation.


News

DSsim 1.1.2

Bug Fixes

  • Fixed bug when running simulations in parallel. Introduced by turning off simulation counter when running in serial.
  • Fixed bug in summary when some simulation iterations fail
  • Fixed detectability plot for discrete covariate distribution with continuous relationship with scale param of the detection function.
  • Now outputs covariate values in addition to other data from data.for.distance function.
  • No longer crashes out of simulation if dht fails it warns, skips dht and continues.
  • Now checks to make sure the user is warned of -ve cluster size values that are generated.

DSsim 1.1.1

Notes

  • Can now turn off the counter when running in serial.
  • Now if the number of spacing values supplied to the systematic line or point transect designs generated by DSsim do not match the number of strata only the first value will be used.

Bug Fixes

  • Corrected covariate distribution documentation
  • Fixed cluster size bugs, there was a problem with the expected cluster size being saved and the true expected cluster size calculations which resulted in error messages.
  • Fixed discrepancy in truncation distance between detectability and analysis - this was introducing some bias. Also now warns the user if the truncation distance for analysis is larger than that for data generation.

DSsim 1.1.0

Notes

  • read.from.file argument in generate.transects() function has been removed. Whether to read from file is now determined by whether the user has supplied a path argument to make.design().

New Features

  • Generates systematic point transect surveys (no need to generate survey shapefiles in Distance in advance).
  • A check.sim.setup function to allow easier checking of the simulation setup.
  • Default values so basic simulations can be run with minimal provision of arguments.
  • Generates systematic line transect surveys (no need to generate survey shapefiles in Distance in advance).
  • A vignette investigating the effects of truncation distance.
  • Individual level covariates at the data generation and model fitting stages.
  • Plotting functionality to show the effects of the covariates on detectability.
  • Implements progress counter when running in serial and progress bar when running in parallel.

Bug Fixes

  • Fixed bug in make.design which was looking to the global environment for an object called region.
  • Ammended RMSE description in description summary

DSsim 1.0.8

New Features

  • Now uses a buffer region for generating the density grid rather than jittering the points. Now imports rgeos and sp.

Bug Fixes

  • If present and if the user has not switched the check.LinkID option in make.region off DSsim uses the LinkID attribute to rearrange the survey strata to match the order they would be displayed in Distance. This is important as otherwise the strata ID's for each transect generated by Distance will be wrong.
  • Fixed the bug that meant that DSsim threw an error when the user tried to create a density grid from a gam. Added testing for this feature.
  • Fixed a bug in plotting density object when there are gaps between polygons in the survey region. Now imports fields.

DSsim 1.0.7

New Features

  • Implements segmented track and segmented grid designs

DSsim 1.0.6

Bug Fixes

  • Fixed RMSE bug with NAs
  • Fixed backwards compatability bug when no analysis strata exists in analysis objects

DSsim 1.0.5

Notes

  • Objects in a population may be detected from any transect. This means objects can be detected from two or more transects.

New Features

  • Now generates point transects including the generation of nested point transect designs (for dual receiver type acoustic surveys)
  • Now implements AICc and BIC model selection criteria in addition to AIC
  • Stores how many times each model was selected based on information criteria and displays this information in the simulation summary.
  • Stores the delta AIC/AICc/BIC to second best model if there is one (however, not currently displayed in summary)
  • save.sim.results will save simulation replicate results to csv files
  • new function to format data ready to be read into Distance (format.for.distance)

Bug Fixes

  • Contours are now switched off in density plot when density is constant
  • Additional shapefile type checking takes place (for region and transect shapefiles)

DSsim 1.0.4

Notes

  • Objects in a population may be detected from any transect. This means objects can be detected from two or more transects.

New Features

  • Uniform probability of detection in population detectability
  • Now accepts all types of suitable ESRI shapefiles
  • make.ddf.analysis.list now can optionally take in a data.frame which describes how strata should be grouped for analysis. See ?make.ddf.analysis.list

Bug Fixes

  • Gaps are now allowed to share a common boundary point with the strata / survey region boundary.

DSsim 1.0.3

Notes

  • Objects in a population may be detected from any transect. This means objects can be detected from two or more transects.
  • Please note that the heat colour have now been inverted by default to be more consistent with normal use. Red is now low and white is high.

New Features

*Better plotting of density surfaces - now displays a grid with contours by default. There is still the option to plot the previous style (points) now with contours option. *There is now the plot option to fill the region polygons and gaps in the polygons with a specified colour.

DSsim 1.0.2

Notes

  • Please note that currently objects in a population may only be detected from the nearest transect. This can sometimes results in edge-like effects if transects are located close together, e.g. with parallel random designs.

New Features

  • Added ability to save and load data generated from simulations
  • Added ability to run binned analyses
  • It now records the number of failed iterations
  • It now checks to see if there is a file meta.txt in the folder with the transects as this is a more robust way of telling it which transects belong to which strata
  • Now performs more checks on the density grid and makes sure that the required population size has been created
  • You can now provide multiple scale and shape parameters to generate a population with individuals of differing detectability
  • You cannow use adjustment terms in the mrds models
  • Added a glossary of terms for the output

Bug Fixes

  • Now catches ddf failures
  • Fixed bug in plot(density) and plot(region)
  • Fixed the way it checks for non-convergence in ddf
  • Fixed NA/NULL bug when ddf failed
  • Fixed bug when some strata have no indivuduals in them
  • Fixed bug when reps = 1
  • Added check for strata names to make sure they are unique
  • Added a function to fix shapefile generated by Distance
  • Updated the documentation

Reference manual

It appears you don't have a PDF plugin for this browser. You can click here to download the reference manual.

install.packages("DSsim")

1.1.2 by Laura Marshall, a year ago


Browse source code at https://github.com/cran/DSsim


Authors: Laura Marshall <[email protected]>


Documentation:   PDF Manual  


GPL (>= 2) license


Imports graphics, splancs, mgcv, shapefiles, rgeos, fields, sp

Depends on mrds, methods

Suggests testthat, parallel, pbapply, knitr, rmarkdown


See at CRAN