StatisticsOp Operator Description

Overview

Name:	`StatisticsOp`
Full name:	`org.esa.beam.statistics.StatisticsOp`
Purpose:	Computes statistics for an arbitrary number of source products. This is the NASA version and slightly varies from the BEAM version. Version set at 7.4.1 (it was 5.0) to not be confused with BEAM versioning scheme.
Version:	7.4.1

Description

An operator that is used to compute statistics for any number of source products, restricted to regions given by an ESRI shapefile.

It writes two different sorts of output:

an ASCII file in tab-separated CSV format, in which the statistics are mapped to the source regions
a shapefile that corresponds to the input shapefile, enriched with the statistics for the regions defined by the shapefile

Note: this operator differs from the GUI version. A key difference being that for this operator:

Currently only one entry is made per band
A shapefile is used instead of a ROI mask.
A math band expression is available
A valid pixel expression is available (the ROI mask could be added here to emulate the GUI version).

Unlike most other operators, that can compute single {@link org.esa.beam.framework.gpf.Tile tiles}, the statistics operator processes all of its source products in its {@link #initialize()} method.

    Note: this operator has been revised as of SeaDAS 7.4.1 to match the SeaDAS 7.4.1 GUI revision of statistics.  The following are the changes to this operator:
    1. Changed field "median" to actually be calculated from the raster median.  Previously it was the 50% threshold of the histogram.
    2. Removed the calling parameter "accuracy", and instead using "numBins".
    3. Default percent thresholds are now {50,80,85,90,95,98}, previously they were {90,95}.
    4. "Max Error" field name changed to "BinWidth"
    5. Added "CoefficientOfVariation" field and "TotalBins" field
    6. Changed all field names and their order to match that of the GUI.

Sources

Name	Description
`sourceProducts`	The source products to be considered for statistics computation. If not given, the parameter 'sourceProductPaths' must be provided.

Parameters

Name	Data Type	Default	Description	Constraints
`sourceProductPaths`	`String[]`		A comma-separated list of file paths specifying the source products. Each path may contain the wildcards '*' (matches recursively any directory), '' (matches any character sequence in path names) and '?' (matches any single character). If, for example, all NetCDF files under /eodata/ shall be considered, use '/eodata/*/.nc'.
`shapefile`	`File`		An ESRI shapefile, providing the considered geographical region(s) given as polygons. If null, all pixels are considered.
`startDate`	`UTC`		The start date. If not given, taken from the 'oldest' source product. Products that have a start date before the start date given by this parameter are not considered.	format: yyyy-MM-dd HH:mm:ss
`endDate`	`UTC`		The end date. If not given, taken from the 'youngest' source product. Products that have an end date after the end date given by this parameter are not considered.	format: yyyy-MM-dd HH:mm:ss
`bandConfigurations`	`BandConfiguration[]`		The band configurations. These configurations determine the input of the operator.	not null
`sourceBandName`	`String`		Within band configurations. Band name	required if no expression given
`expression`	`String`		Within band configurations. Math band expression (can be used instead of band name)	required if no band name given
`validPixelExpression`	`String`	`Existing valid pixel expression of band`	Within band configurations. Valid pixel expression for band.
`outputShapefile`	`File`		The target file for shapefile output. Shapefile output will only be written if this parameter is set. The band mapping file will have the suffix _band_mapping.txt.
`outputAsciiFile`	`File`		The target file for ASCII output.The metadata file will have the suffix _metadata.txt. ASCII output will only be written if this parameter is set.
`percentiles`	`int[]`	`50,80,85,90,95,98`	The percentile levels that shall be created. Must be in the interval [0..100]
`numBins`	`int`	`1000`	The number of bins to be used in the histogram statistics.