Regrid

Regrid is a tool for creating new data sets by manipulating the data either in pixel space or in metadata space. Regrid will manage up to two datasets at a time. If two data sets are dropped in they can be registered to each other or to any arbitrary grid supplied.

In pixel space the data can be manipulated to make a new data set in the following ways:
(1) using every n^th value,
(2) limiting the data to within a rectangular region, or
(3) resampling to an arbitrary size.

In metadata space the data can be manipulated to make a new data set in the following ways:
(1) selecting new endpoints for the original metadata to assume.
(2) resampling to an arbitrary size.
(3) registering (i.e. matching up geographic boundaries) one dataset to the other.

These (when logical) can be combined.

Only datasets that have explicit or implied geo-referencing (i.e. geographic) information can be registered. Examples of explicitly geo-referenced data sets are HDF-EOS point, grid and swath files and PDS files that have projection information. An example of an implied geo-referenced dataset is a raw binary file where the metadata is supplied by the user and the metadata labels are "Latitude" and "Longitude".

Registering two datasets that are projected will result in the destination datasets projection information being used by the source dataset. (e.g. if it is desired that a sinusoidal projected data be registered to Lambert projected data, the data in the sinusoidal projection will be reprojected into the Lambert projection.)

If two geo-referenced datasets are registered but there are no common geographic areas, then no data will appear for the dataset whose metadata is outside the selected region.

Regrid is useful when a data set is very large and the region of interest in it is small. Its output is available through the drag and drop button(s) that will appear once the "Accept" button is pressed.

Controls

Buttons

The Accept button creates a new database given the current settings in the text pads.

The Reset button resets input values to the original values from the dataset that was dropped in.

Menus

The Interpolation menu allows one of two choices of interpolation schemes. "Nearest Pixel" means that the closest old data point to the center of the new data point is chosen. "Smooth Interpolation" means that the new data point is a weighted average of the old data points along that particular axis. Smooth interpolation ONLY works for forced resampling; normal decimation uses "nearest pixel." So, if you want to halve the width of your dataset by averaging each pair of data points together, you must use the forced resampling method, rather than decimation.

The Menu button allows the user to switch between resampling in pixel space and resampling in metadata space. If there have been two datasets dropped in and they are recognized as being able to be registered then the menu will present options to register one dataset to the other.

Sliders

Sliders control the edges of the rectangular portion of the database to use. The "Start:", "End:", and "Resample:" text fields update as the slider for the chosen axis is moved. The meta coordinates of the start and end of the new dataset are displayed next to the axis name.

Text Input

Pixel space mode:

The "Start:" and "End:" fields set the limits of the data. The "Step:" field is used for decimation; a value of n here means every nth value of the old data set will be mapped into the new data set. The "Resample:" fields allow the user to set what the dimensions of each axis should assume.

Metadata mode:

The "Start:" and "End:" fields set the limits which the new metadata should assume. The "Resample:" fields allow the user to set what the dimensions of each axis should assume.

Applications to Connect to This Application

Any Data Object may be connected via the "drag 'n drop" button to this application. An application with a bounding box (other than Scat2D) will set the sliders.

Applications to Connect This Application to

Regrid may be connected via the "drag 'n drop" button to Slider, RGB Slider, Combine Slider, 3-Slider, PanZoom, LightTool, 2AxisRotator, 3AxisRotator, Image, Window Tool, Plane, Globe, Histogram, Line Plot, Average, Combine, Compare, Profile, FFT, 2-D Scatterplot, Track Pixel, Value View, Calculator, and Contour.

Tricks and Gotchas

Resampling occurs one axis at a time in order. In other words, if you have a 10x20x30x40x50 data set to resample to 5x6x7x8x9, a series of intermediate datasets of size 5x20x30x40x50, 5x6x30x40x50, 5x6x7x40x50, and 5x6x7x8x50 are created.

If you decimate an axis, remember that the values in the "Resample:" boxes are the final size of the new data set, so set them to the values you want.

Up to two datasets can be managed at one time. This is especially useful for data that have overlapping geographic regions, but may have different resolutions.

If you are subsetting or resampling a dataset that has projection information and the resolution of the data is changed so that the projection parameters no longer have meaning (e.g. resampling axes so the resolution on the x-axis is different than the resolution on the y-axis) there will not be any projection information supplied in the output dataset.

If you have a 4D dataset, the 4th axis sliders/text pads can be used to specify a slice on the 4th axis. If you don't choose a slice this way, you will see only the first slice of this 4th axis.

WebWinds Home / Nov 7, 2001