Learning goals

  • Image properties, image encoding, raster formats and metadata in remote sensing
  • Get in touch with airborne imaging spectroscopy (hyperspectral) data
  • Introduction and first steps with the EnMAP-Box
  • How to develop image spectral libraries

Background

Image properties

The most prominent properties of digital remote sensing imagery include the number of spectral bands, spatial resolution, radiometric resolution and the image dimension.

  • The number of spectral bands represents the number of spectral measurements taken in the range of wavelengths as a function of a sensor’s spectral resolution and spectral sampling interval (see Session 4).
  • The spatial resolution describes the length of one side of a single image pixel. Often, the term ground sample distance (GSD) is used interchangeably, which describes the distance between two adjacent pixel centers measured on the ground.
  • The radiometric resolution describes the range and discernible number of discrete gray values a sensor can distinguish. The radiometric resolution is measured in bit.
  • The image dimension is defined by the number of columns (samples) and number of rows (lines) of an image.

The memory needed to store an image can be calculated as follows:

\(image size = number of bands \times number of columns * numbers of rows * bit width\)

Properties of digital remote sensing imagery (Source: Richards, 2006)

Data types

Data types specify the values digital imagery may contain and constrain the values that an expression, such as a variable or a function, may take. Common data types of digital remote sensing imagery are illustrated in the table below.

Data type Typical bit width Value range
byte 1 byte = 8 bit -127 to 127 (signed); 0 to 255 (unsigned)
integer 2 byte = 16 bit -32768 to 32768 (signed); 0 to 65535 (unsigned)
float 4 byte = 32 bit -Inf to +Inf, ‘single precision’; floating point number according to IEEE 754
double 8 byte = 64 bit -Inf to +Inf, ‘double precision’

Storage structure

Band interleaved by pixel (BIP), band interleaved by line (BIL) and band sequential (BSQ) are common approaches to organize multiband images. BIP, BIL, and BSQ are not image formats themselves but structures for storing the gray values of an image in memory or on disk.

  • BIP: for each band, gray values are stored in a pixel-wise manner
  • BIL: bands are stored line-wise
  • BSQ: entire bands are stored on disk one after the other

BIP, BIL and BSQ storage structure examplified for an image with three bands of n pixels each

Image formats

There are hundreds of different raster image formats (see gdal.org), most common image formats regarding digital remote sensing imagery are illustrated in the table below.

Format name Extension Description
GeoTiff .tif, .tiff, .gtiff TIFF + geospatial reference
ENVI , .bsq, .bil, .bip, .dat generic, often used in imaging spectroscopy community; Header file (.hdr) with meta data!
JPEG2000 .jp2, .j2k used by many data providers; usually for integer values only
HDF4, HDF5 .hdf, .h4, .hdf4, .h5, .hdf5 hierarchical data format, version 4 or 5; multi-resolution raster
netCDF Network Common Data Forat; multi-resolution raster
SAVE Standard Archive Format for Europe e.g. Sentinel-1 and Sentinel-2

Metadata

Metadata are information about the data and are commonly automatically recorded during data acquisition or added to the data during pre-processing. Metadata are either stored as part of an integrated data format, i.e., stored in the same file, or as an additional file accompanying the image data.

  • Example GeoTIFF: metadata stored within .tif image. Displaying metadata possible with specific metadata reader (e.g. via QGIS Layer properties).

  • Example ENVI format: metadata stored in separate header file (.hdr). Displaying and editing metadata possible with text editor.

HyMap data Berlin

HyMap (Hyperspectral Mapper) is an airborne imaging spectrometer covering the spectral range from 450 to 2500 nm in 128 bands with spectral resolutions between 10 and 20 nm. The spatial resolution varies with operating flight altitude (i.e., 2000 – 5000 m above ground level) and is usually in the range between 2 and 10 m.

HyMap sensor (Source: NASA) and HyMap data recored along an east-west gradient of Berlin

HyMap imagery are used in various application fields, including detailed mapping of urban land cover. The high spectral (hyperspectral) resolution of HyMap data enables the differentiation of surface cover types which are not distinguishable in broadband multispectral data. Due to the relatively high spatial resolution, HyMap imagery can be used as alternative source to lab- and field-spectroscopy for developing spectral libraries of different materials and natural surface cover types.

Subset of Berlin, Charlottenburg, as represented in VHR Google Satellite imagery and HyMap imagery

Spectral library of urban surface spectra ordered by land cover type as extracted from an HyMap image at 3.6 m resolution from Berlin, Germany (Source: van der Linden et al. 2018)

EnMAP-Box

The EnMAP-Box is developed at the Humboldt-Universität zu Berlin and the Universität Greifswald as a free and open source plug-in for QGIS. The EnMAP-Box is designed to process imaging spectroscopy data and particularly developed to handle data from the upcoming hyperspectral satellite mission Environmental Mapping and Analysis Program (EnMAP). The EnMAP Box enables you to:

  • Extend your QGIS for remote sensing image analysis
  • Add powerful tools to process and analyse imaging spectroscopy data
  • Integrate machine learning algorithms into your image classification and regression with Random Forests, Support Vector Machines and many more
  • Create and manage spectral libraries with attribute data
  • Develop your own image processing algorithms using a powerful Python API

More information about the EnMAP-Box, including installation guidelines, user manuals or different application tutorials can be found on the EnMAP-Box project page.

Interface of the EnMAP plugin in QGIS

Session materials

Download the session materials from our shared repository. The materials contain a subset of a HyMap scene from Berlin/Brandenburg, which was extracted from Berlin-Urban-Gradient dataset.

Exercise

Sentinel-2 & HyMap metadata

  • Open the header files (.hdr) of the Sentinel-2 image (Session 2: ‘Sentinel2_T33UUU_20190726_10m_4bands_subset_berlin.hdr’) and the HyMap image (‘hymap_20090820_4m_111bands_subset_berlin.hdr’) in a text editor (e.g., Editor or Notepad++)
  • Which meta-information can you find in the header files regarding:
    • Image dimension
    • Spatial resolution
    • Number of bands
    • Radiometric resolution
    • Data type
    • Storage structure
  • Which additional meta-information is provided with the headerfiles?

First steps in the EnMAP-Box

This video (includes sound) illustrates the basic functionality for visualizing remote sensing imagery in the EnMAP-Box.

Assignment

The aim of this assignment is to develop a spectral library based the HyMap data from Berlin and the spectral viewer functionality of the EnMAP-Box.

Develop a spectral library

  • For each of the classes below, collect 4 reflectance spectra from spectrally pure surfaces in the HyMap image:
    • Building
    • Sealed non built-up
    • Coniferous forest
    • Deciduous forest
    • Grass
    • Bare land (soil/sand)
    • Water
  • Add two attribute columns (‘class’, ‘description’) and provide the corresponding class (see table above) and a detailed description of the material / surface (e.g., ‘red clay tile’).
  • Style (colorize) the spectra appropriately with regard to the classes.
  • Save the collected spectra as a Spectral Library file (.sli).
  • Export an image of your spectral library (including a legend).

How to develop a spectral library in the EnMAP-Box

Submission

  • Create a zip-archive containing the image of your spectral library as well as the saved spectral library (.sli + .hdr) and upload the zip-file on Moodle.

  • General submission notes: Submission deadline for the weekly assignment is always the following Monday at 10am. Please use the naming convention indicating session number and family name of all students in the respective team, e.g. ‘s01_surname1_surname2_surname3_surname4.pdf’. Each team member has to upload the assignment individually. Provide single file submissions, in case you have to submit multiple files, create a *.zip archive.


Copyright © 2023 Humboldt-Universität zu Berlin. Department of Geography.