Focal PCA on a Grid

Author(s)	A.Thomas, O.Conrad (c) 2016
Library ID	statistics_grid
Tool ID	15
Version	1.0
Menu	Spatial and Geostatistics \| Grids \| Principal Components

Description

This tool uses the difference in cell values of a center cell and its neighbours (as specified by the kernel) as features for a Principal Component Analysis (PCA). This method has been used by Thomas and Herzfeld (2004) to parameterize the topography for a subsequent regionalization of climate variables with the principal components as predictors in a regression model.

References

Benichou, P., Lebreton, O. (1987): Prise en compte de la topographie pour la cartographie des champs pluviometriques statistiques. Meteorologie 7. Serie, no. 19.

Thomas, A., Herzfeld, U.C. (2004): REGEOTOP: New Climatic Data Fields for East Asia Based on Localized Relief Information and Geostatistical Methods. International Journal of Climatology, 24(10), 1283-1306. DOI:10.1002/joc.1058. Wiley Online Library

Parameters

	Name	Type	Identifier	Description	Constraints
Input	Grid	grid, input	`GRID`	-	-
Output	Base Topographies	grid list, output, optional	`BASE`	-	-
	Principal Components	grid list, output	`PCA`	-	-
	Eigen Vectors	table, output, optional	`EIGEN`	-	-
Options	Grid System	grid system	`PARAMETERS_GRID_SYSTEM`	-	-
	Number of Components	integer number	`COMPONENTS`	number of first components in the output; set to zero to get all	Minimum: 1 Default: 7
	Output of Base Topographies	boolean	`BASE_OUT`	-	Default: 0
	Overwrite Previous Results	boolean	`OVERWRITE`	-	Default: 1
	Kernel Type	choice	`KERNEL_TYPE`	-	Available Choices: [0] Square [1] Circle Default: 1
	Kernel Radius	integer number	`KERNEL_RADIUS`	Kernel radius in cells.	Minimum: 1 Default: 2
	Method	choice	`METHOD`	-	Available Choices: [0] correlation matrix [1] variance-covariance matrix [2] sums-of-squares-and-cross-products matrix Default: 1

Command Line


Usage: saga_cmd statistics_grid 15 [-GRID ] [-BASE ] [-PCA ] [-EIGEN ] [-COMPONENTS ] [-BASE_OUT ] [-OVERWRITE ] [-KERNEL_TYPE ] [-KERNEL_RADIUS ] [-METHOD ]
  -GRID:            Grid
	grid, input
  -BASE:            Base Topographies
	grid list, output, optional
  -PCA:             Principal Components
	grid list, output
  -EIGEN:           Eigen Vectors
	table, output, optional
  -COMPONENTS:      Number of Components
	integer number
	Minimum: 1
	Default: 7
  -BASE_OUT:        Output of Base Topographies
	boolean
	Default: 0
  -OVERWRITE:       Overwrite Previous Results
	boolean
	Default: 1
  -KERNEL_TYPE:     Kernel Type
	choice
	Available Choices:
	[0] Square
	[1] Circle
	Default: 1
  -KERNEL_RADIUS:   Kernel Radius
	integer number
	Minimum: 1
	Default: 2
  -METHOD:
                         Method
                         	choice
                         	Available Choices:
                         	[0] correlation matrix
                         	[1] variance-covariance matrix
                         	[2] sums-of-squares-and-cross-products matrix
                         	Default: 1