stateVector

Design and build state vector ensembles


Description

The state vector class is used to design and build state vector ensembles. The class implements an object that describes how a state vector ensemble should look. Thus, the class allows users to design a state vector ensemble without manipulating any data. Once a user finishes designing a state vector ensemble, they can call the “build” command to automatically build the state vector ensemble from a set of gridfile data catalogues.

Concepts

The following is some vocabulary useful for designing state vector ensembles. Ensemble Dimensions are the data dimensions used to select different ensemble members - often, these are the time and run dimensions. State Dimensions are the data dimensions that are organized down the state vector. These are often the spatial dimensions used to describe the dataset. In general, state dimensions will have the same metadata value at a state vector element, regardless of the ensemble member. By contrast, ensemble dimensions will have different values for different ensemble members.

Sometimes, an ensemble dimension will also have elements down the state vector (for example, successive months of the year). This is referred to as a sequence. The term coupled variables indicates that, within an ensemble member, the variables should be selected from the same points along the ensemble dimensions (for example, from the same points in time or from the same climate model run). In most cases, all the variables in an ensemble should be coupled to one another.

Outline

The following is a rough sketch for using the stateVector class.

  1. Use the “stateVector” command to initialize the design for a state vector.

  2. Use the “add” command to add variables to the state vector

  3. Use the “design” command to indicate the ensemble dimensions, and to select subsets of gridfile catalogues that should be used to build the state vector ensemble.

  4. Use the “sequence” command to implement sequences

  5. Use the “mean”, “total”, “weightedMean”, and “weightedTotal” commands to implement means and totals in the state vector.

  6. Use “build” to generate a state vector ensemble from the design.

The class also includes various methods that return information about the state vector. Among others, the “variables”, “length”, “members”, and “label” commands can all help facilitate workflows.

Troubleshooting Metadata

Sometimes, variables have different metadata formats along the ensemble dimensions (for example, annual and monthly time metadata). This can prevent stateVector from building an ensemble, because the class will not be able to check that the values in different ensemble members align to the same points. If this occurs, use the “metadata” command to allow comparison of the variables with different metadata formats.


Key Methods

The following methods are the most essential for users.


All User Methods

The complete list of methods for users.

Create

Variables

Design

Metadata

Overlap

Coupling

Build

Size Information

Variable Information

General Information

Serialization


Utility Methods

Utility methods that help the class run. They are not intended for users.

General

Variables

Indices

Coupling

Build

Console Display

Gridfile Interactions

Unit Tests