The code here is covered by the GNU General Public License https://www.gnu.org/licenses/gpl.html
Tetracorder is a spectral identification and mapping system designed for mapping materials on solid surfaces in the Solar System, including the Earth, other planets and satellites, asteroids and comets. Tetracorder can also be used in laboratory spectroscopy. Tetracorder has 2 modes: 1) identification of components in a single spectrum, and 2) identification and spatial mapping using imaging spectrometer data.
Specpr (Spectrum Processing Routines) is a spectral analysis system for analyzing single spectra. Tetracorder uses many specpr subroutines and it is required to be compiled before tetracorder.
Training Videos are available on the Planetary Science Institute youtube channel: https://www.youtube.com/user/PSITucson/featured. Go down to "Spectroscopy and Tetracorder Training" and click on "PLAY ALL" to see all the training videos (listed on the right).
The source code is located in the tetracorder5.26 and specpr directories.
The required spectral libraries are in the sl1 directory and includes code and instructions to convolve the spectral library to other instruments.
The etc directory contains environment variable definitions for bash and tcsh.
The tetracorder.cmds directory includes the tetracorder expert system that does the identification, and all the files needed to do an analysis on a dataset. For the system to operate, the spectral libraries must be convolved to the spectral range and resolution of the instrument supplying the data, including files with pointers to the convolved libraries so that tetracorder knows which libraries to use for a given instrument.
The cuprite95 directory contains the results from a tetracorder run on NASA AVIRIS cuprite 1995 data that was calibrated to apparent surface reflectance. You can get the image cube from the USGS spectroscopy lab ftp site (see the README-image-cube.txt file in the cuprite95 directory for the location). Then you can run tetracorder and confirm that you get the same results. See the training videos for how to evaluate results.
The tetracorder system also requires Davinci, available from Arizona State U here
Tetracorder, Specpr, and support programs run on linux and unix.
Github does not have the languages right. The specpr and tetracorder programs are fortran, ratfor, and C. Support programs are mostly davinci and shell scripts, with some fortran/ratfor and C programs. HTML is documentation in the spectral libraries. Spectral libraries are binary data.
Detection and mapping of minerals, vegetation species, chemicals, liquids, and solids is being done through the field of imaging spectroscopy. Imaging spectrometers are deployed on aircraft, spacecraft (throughout the Solar System) in the field, and in laboratories. Imaging spectrometers are narrow-band imagers with hundreds of wavelengths (bands) and usually include ultraviolet to infrared. Imaging spectrometers collect a spectrum at each image pixel with enough spectral resolution to resolve absorption and emission features in spectra of compounds, whether crystalline or amorphous solids, liquids and gases. What can be detected remotely is mainly a function of spectral range and resolution. Analysis of imaging spectrometer data sets is quite complex. The Tetracorder system was first described in detail in Clark, R.N., Swayze, G.A., Livo, K.E., Kokaly, R.F., Sutley, S.J., Dalton, J.B., McDougal, R.R., and Gent, C.A., 2003, Imaging spectroscopy: Earth and planetary remote sensing with the USGS Tetracorder and expert systems, Journal of Geophysical Research, Vol. 108(E12), 5131, doi:10.1029/2002JE001847, p. 5-1 to 5-44.
Tetracorder is in use analyzing data from all over the Solar System, including mapping ice and other compounds on icy satellite surfaces in the Saturn and Jupiter systems, minerals on Mars, and was critical in making the discovery of widespread water on the Moon possible. Tetracorder is used for mapping ecosystems, and in rapid response to environmental disasters. It was used in assessing the environmental damage from the World Trade Center disaster and the the 2010 Deepwater Horizon oil spill in the Gulf of Mexico.
Tetracorder uses multiple algorithms, including spectral fitting procedures to identify materials, and derives feature strengths (relative abundance) of those materials. It has demonstrated discrimination of grain sizes for some materials using the shape of the absorption features. The grain sizes and feature strengths can then be fed into radiative transfer models to derive component abundances.
Tetracorder 5.x+ adapts to both environmental conditions as well as instrument capability. Previous versions had to be adapted by an expert spectroscopist for each sensor and environment the data came from. Tetracorder produces maps of hundreds of materials, including chemical substitutions in some minerals. Imagine doing chemistry remotely, whether across a canyon, from a high point overlooking an environmental disaster, or a studying remote planets. This is now possible. Further, Tetracorder results can be made into custom color-coded maps automatically.
Tetracorder is the mineral identification used used for the NASA Earth Surface Mineral Dust Source Investigation (EMIT) imaging spectrometer instrument that will go on the International Space Station in June 2022. EMIT will determine the mineral composition of natural sources that produce dust aerosols around the world. By measuring in detail which minerals make up the dust, EMIT will help to answer the essential question of whether this type of aerosol warms or cools the atmosphere.
June 2022: Tetracorder 5.27 released. The 5.27 code introduces a new spectral feature class, class M. Class D, diagnostic, could still find a material based on other features if a diagnostic feature was disabled. The result may not be correct. The condition was discovered in AVIRIS data when the reflectance calibration had a large deleted zone around the 1.4 and 1.9-micron telluric water bands. For example, in group 1, the right continuum point for azurite (copper carbonate) was inside the deleted channel zone and the main diagnostic feature was disabled. Identification fell to a less diagnostic and less unique feature resulting in widespread mapping of azurite, but that mapping was a false positive. Tetracorder 5.27 with the M spectral class is Must Have Diagnostic feature and if the feature is disabled, the material will not be found. The 5.27 expert system has been updated where multiple materials now use the class M diagnostic. Testing in multiple geologic environments shows the false positive rate is vastly reduced.
A study was conducted on the ID of snow+vegetation and the expert system was adjusted. Some plants have shifted water bands that are similar to those in snow+vegetation spectra. The false positive rate for snow+vegetation is now reduced. If you know the temperature range of your scene, setting the temperature will help reduce the false positive snow detection if temperatures are above freezing.
Tetracorder 5.27 includes additional spectra for mapping materials in the 3 to 4-micron range. Specifically weak carbonate signatures are now included.
The color map products for the 2-micron spectral features have also been improved with better color to distinguish minerals better. A map of muscovite composition has been added.
January 2023: Minor updates to Tetracorder 5.27, mostly debugging and printing information for single-spectrum mode including improvements for audio output for what tetracorder finds in a spectrum. Before audio output is effective, the wav files for each found material needs to be updated (planned for this year).
This update also expanded the number of spectral groups and the spectral range out to 5 microns with definitions to go to 20+ microns. To do this, the group arrays needed to be increased and that made tetracorder go beyond the small memory model in linux. The linux memory model has a 2-gigabyte limit in default compile mode and I hit that. Before crossing that limit with new compiler flags and new memory model, I want to to do some performance testing below and above the limit. In the meantime, I reduced some array sizes that were far larger than needed, so everything fits just fine.
The 5.27c1 expert system in the tetracorder.cmds directory has been expanded to 5 microns, though more population of reference features is needed. Also many improvements were made based on mapping with the first EMIT data from the International Space Station, and the NASA SSERVI TREX field campaigns from October 2022, including results from the Carnegie Melon Zoe rover which is running tetracorder (with Ubuntu Linux) providing real-time analysis from a visible-near-IR spectrometer.
BE SURE TO READ THE AAAAA.KNOWN-ISSUES.txt file:
t1/tetracorder.cmds/tetracorder5.27c.cmds/AAAAA.KNOWN-ISSUES.txt
And if you find errors or ways to improve the expert system, please let me know.
For the 5.27c1 improvements, several spectra were added to the sprlb06a spectral library. That meant the spectral libraries needed to be be re-convolved. Rather than do updates that can take extra space, I have elected to try a new method: delete the old libraries from github and install the new. The new libraries are completely compatible will all older versions of tetracorder and expert systems, because we only add new spectra and never modify existing ones.
April 26, 2023: A condition was found when mapping an EMIT scene in Morocco, and a larger than plausible amount of jarosite and gypsum. See the tetracorder.cmds/tetracorder5.27c.cmds/AAAAA.KNOWN-ISSUES.txt file for details. A mitigation study is underway.
NOTE on adding entried to the spectral libraries. When new entries are added to the spectral libraries, the convolved libraries need to be regenerated. There is no need to store changes, as only additiyons are added and they are always fully backward compatible. Thus to update the libraries, the entire library directory is removed from github and the new one added. This prevents huge data volume growth that is not needed.
April 28, 2023: updated the rlib06 library and convolved libraries by adding 3 new spectra.