[Index] [Prev] [Next] [Bottom]


Chapter 3: Opening HDF Data Objects

HDF is a library and platform independent data format for the storage and exchange of scientific data. HDF files may contain many types of information that a scientist might need.

JHV can be used to view several types of HDF data objects. Due the differing natures of these types of data sets, the manner in which they open for display differs from one type to another. For example, an 8-bit raster image will open a preview copy of the image, a 3D SDS first asks how the data is to be sliced for viewing, and a Vdata opens as a data table. This chapter describes opening each of the data types in a JHV session.

Currently JHV can be used with five types of HDF objects:

3.1 File and Data Object Annotations
3.2 Raster Images
3.3 Scientific Data Sets
3.4 Vdatas, ot Tables
3.5 Vgroups

The last four types of data above are the essential data sets; HDF was created to store these types of information in a widely-accessible format. Annotations, the first item, are a fundamentally different sort of data; they exist solely to describe the data stored in the HDF file.

Complete information regarding these data structures is available from The HDF Home Page.


3.1 File and Data Object Annotations

HDF annotations are text strings used to describe a file or any of the data elements the HDF file contains.

Annotations come in two forms; labels and descriptions. Labels are short annotations used for assigning things like titles or time stamps to a file or its data objects. Descriptions are longer annotations; they typically contain more extensive information, such as a source code module or mathematical formulae.

There are also two types of annotations: file annotations and data object annotations. A file annotation describes an entire HDF file. A data object annotation describes a particular data object within the file.

File and data object annotations can each be either labels or descriptions. Thus an annotation might be a file label, a file description, an object label, or an object description.


Figure 3.1 A file description, one sort of annotation, displayed in the message window

3.2 Raster Images

The general raster, or GR, data model is designed to provide a flexible means of storing raster image data, including 8-bit raster or 24-bit raster images. Raster image data is stored in a two-dimensional array and attributes may be associated with the image, the file, or both. Palettes (also known as CLUTs or color look-up tables) and compression method information can also be associated with an image.

For backward compatibility, the HDF library allows users to access images as a generic raster image or as an 8-bit or 24-bit raster image. In such a case, the image is stored once, but under two identities; JHV will show two objects in the hierarchy, although there is really only one image in the file.

GR object annotation
A GR object annotation contains information about the image. The information can be displayed in the text area by clicking the object folder. The following is an example of a GR object annotation.

 
  Image Name : Raster Image #1
  Image Index: 1
  Image Size : 625 by 271
  Data Number Type: 8-bit unsigned char
  Num of Components: 1
  Interlace  : 0
  Num of Attributes: 0
  

GR data attributes
Attributes are defined by the calling program to contain auxiliary information about a file, an image, or both. Click the attribute object icon in the hierarchical tree to view the data attributes. The data attributes are displayed in Name/Value format in the message window.


Figure 3.2 General raster image attributes displayed in the message window
GR preview image
JHV has a limited image display capability. When you select an image, JHV will first present a compressed preview of the image in the fixed-size preview image window. You can then select subsamples of the image for further display. Note that the entire image is displayed in the preview window when the image is the same size or smaller than the window.

To examine a portion of the image at greater resolution (up to the resolution of the original image), drag the mouse on the preview image to select a subsample or click on the image to select the full image. The selected subsample will be displayed in a separate image window.


Figure 3.3 GR preview image in the main window (upper panels); selected full image and zoomed subset in the image window (lower panels and described in a later chapter)

3.3 Scientific Data Set

The scientific data set model, or SDS, supports four primary data objects: arrays, dimensions, dimension scales, and dimension attributes. The fundamental object of the model is the SDS array; the remaining objects describe the array.

JHV can display 2- or 3-dimensional SDS arrays. A 2D array is displayed as a single image; a 3D array is displayed as an animated sequence of 2D images sliced along the third axis. See Chapter 7, Animations, for further information.

SDS object information
SDS object information is a description of the SDS object data. Click the SDS object folder to view the object information. Below is an example of SDS object information.

  Dataset name         : Data-Set-2
  Rank                 : 3, 128 X 128 X 65
  Data number type     : 16-bit signed integer
  Number of Attributes : 6
  

SDS data attributes
Attributes describe the nature and/or intended usage of the object they are attached to. The object may be a file, an SDS, or a dimension. JHV displays the data attributes in Name/Value format in the message window.


Figure 3.4 SDS attributes displayed in the message window
SDS preview image
An SDS preview image is a compressed image created from SDS array.

A 2D SDS array is displayed as a 2D image in the preview image panel.

A 3D SDS array is displayed as a sequence of 2D images; each 2D image is a slice of the 3D image taken along an axis selected by the user. These 2D images can be viewed singly or as an animated sequence. See Chapter 7, Animations, for further information.


Figure 2.5 SDS Preview Image

3.4 Vdatas, or Tables

An HDF Vdata object is a collection of records whose values are stored in fixed-length fields. Thus, a vdata is like a table in which each row has the same structure and each column contains data of the same type. The Vdata model allows multiple entries per field; the number of entries or components in a field is called the order of the field.
Vdata information
Information regarding a vdata object includes record number, field name, field order, field type, and so forth. This information is displayed in the JHV message window when you click on the vdata object folder. The following is an example of such vdata information:

 
  VStag         : 1963
  ref           : 2
  Vdata Class   : Data
  Vdata Name    : Sea Ice Motion Vectors - 17766010
  Vdata records : 291
  Interlace     : 0
  Field number  : 8
  Field list    : REFLAT,REFLON,DSPLAT,DSPLON,DELTAX,DELTAY,ROTANGLE,MATCH
  Record size   : 45 bytes
  Field name     Field order    Field type
  REFLAT         1              64-bit floating point
  REFLON         1              64-bit floating point
  DSPLAT         1              64-bit floating point
  DSPLON         1              64-bit floating point
  DELTAX         1              32-bit floating point
  DELTAY         1              32-bit floating point
  ROTANGLE       1              32-bit floating point
  MATCH          1              8-bit unsigned integer
  

Vdata table
JHV displays a Vdata table as a spreadsheet with columns and rows representing the Vdata fields and records respectively. The JHV spreadsheet functionality is somewhat limited, supporting only plot and save functions.


Figure 3.6 A Vdata spreadsheet

3.5 Vgroups

An HDF vgroup object is a structure designed to associate related objects. The general structure of Vgroups is similar to the Unix file system of directories, subdirectories, and files in that a Vgroup may contain references to other Vgroups and/or to data objects. Any HDF object can be included within a Vgroup.


Figure 3.7 A Vgroup
Vgroup information
Vgroup information is displayed by clicking on a Vgroup folder in the displayed file hierarchy. The following example shows that the current HDF file contains two Vgroups, AVHRR 1KM Global 10 day Composite and avhrr-ndvi-060192.hdf.

 
  Tag/Ref: 1965/2
  Name: AVHRR 1KM Global 10 day Composite
  Class Name: PlanetaryGrid
  Entries: 2
  Tag/Ref: 1965/42
  Name: avhrr-ndvi-060192.hdf
  Class Name: CDF0.0
  Entries: 5
  

Vgroup data
A Vgroup data object is usually a Vdata or an SDS data set. Clicking on a Vgroup object node opens the Vgroup object in the appropriate vdata or sds mode.


[Index] [Prev] [Next] [Bottom]


NCSA
The National Center for Supercomputing Applications

University of Illinois at Urbana-Champaign

hdfhelp@ncsa.uiuc.edu