PSM
Classes and functions that facilitate using forward models to estimate proxy observations
Description
The PSM package facilitates calculating proxy observation estimates. The package allows users to apply proxy forward models to state vector ensembles in order to estimate proxy values. The package has several main uses. In brief, the PSM package:
Provides classes that implement proxy forward models
Downloads the codebases of supported models from Github, and
Applies forward models to state vector ensembles.
At its core, the PSM package provides an interface that allows users to run a variety of proxy forward models using the same set of commands. You can learn about the commen set of commands by reading the documentation for PSM.Interface.
Outline
The following is a outline for using the PSM package.
Use “info” to learn about the forward models supported by DASH
Use “download” to download the code for the desired forward models
Use the forward model interface to design a forward model for each of the proxy observation sites. a. Use PSM.<model name> to create a new PSM object for a site b. Optionally use “label” to label the forward model c. Use “rows” to indicate which state vector rows hold the inputs for the forward model
Use “estimate” to apply all the forward models to a state vector ensemble and produce proxy observation estimates.
Supported Forward Models
DASH officially supports the Bay* suite of foraminiferal forward models, general linear forward models, the PRYSM suite of Python forward models, and the Vaganov-Shashkin Lite model of tree ring width. See the documentation on the forward model classes listed below for additional details about these models.
Developing New Forward Models
DASH anticipates that many users will want to develop and use additional forward models for data assimilation. To this end, the PSM package provides a template file that allows users to develop new forward models without needing to manipulate the internal code of the DASH toolbox. See “PSM.template” and read the documentation within for help adding new PSMs to DASH.
We welcome contributions! If you would like DASH to officially support a new forward model, please send an email to DASH.toolbox@gmail.com, or submit a pull request to the DASH Github repository.