PDL is useful in many fields, and has a diverse user community. Below is a map of PDL users, as well as some ineresting work done with PDL around the world.
Some PDL success stories
- Astronomy and Astrophysics
- Meteorology and Terrestrial Monitoring
- Medical imaging
Artificial vision of magnetic fields
I and my graduate student use PDL to track thousands of magnetic features on the Sun. The churning surface of the Sun is covered by magnetic north and south poles that are constantly emerging, moving, recombining, and dissipating. Special telescopes called magnetographs can view these poles as black and white spots, allowing us to probe the Sun's dynamo. PDL's Transform package made it easy for us to remove perspective effects from spacecraft data, and to write an artificial vision application to identify every magnetic feature in the field of view.
What's more, PDL is free -- so we can distribute our software for anyone to use.
--Craig DeForest, deforest (at) boulder (dot) swri (dot) edu
Tracked magnetogram shows about 1/10 of the surface of the Sun, reprojected to appear as if you're hovering immediately over it (click for movie). Our software has identified and numbered every statistically significant feature.
Here, poles that appear in N/S pairs are marked by yellow highlights. Single north or south pole appearances (in apparent but not real violation of physics) are marked by blue highlights. Flux is emerging on scales too small for the telescope to fully resolve it.
High speed galactic winds
I did much of the data analysis for my Ph.D. thesis using PDL, and got an AAS press release out of it. I never imagined being able to manipulate 3-D radio science datacubes in Perl until I met PDL. Thanks!
This optical image of the Whirlpool Galaxy (M51) shows the regions of high speed galactic winds that I discovered.
--Eric Miller ( milleric (at) space (dot) mit (dot) edu )
Star Formation throughout the Universe
As a professional astrophysicist, I have analyzed countless types of data using PDL. Here is one example: In 1997 I collected some very deep-field images of the Hubble Deep Field using the Taurus Tunable Filter and the 4m William Herschel Telescope in La Palma. The tunable filter produces 3-D data (x,y,wavelength) that are difficult to analyze with conventional astronomical software. I reduced my data entirely with home-brew PDL scripts that identified galaxies and separated them by wavelength and position. The slicing, dicing, rapid prototyping, and "glue" features of PDL and Perl were all used to the max.
This figure is from a paper that I and some colleagues published in the Astronomical Journal, showing emission line galaxies that we found in the data cube (using another PDL script). Such galaxies have excess emission at one wavelength due to nebular emission from young stars creating an ionized nebula. Each row shows a single galaxy at several different wavelengths, and the graph to the right shows the total brightness of the galaxy at each wavelength. These galaxies are at distances of 5,000 - 10,000 million light years. We used these data to estimate the formation rate of stars throughout much of the history of the Universe.
One interesting fact: the original PDL script used to make this figure (on a Sparc 20 workstation) in 1998 still worked unchanged (on a G5 Mac) in 2004 when the final paper was submitted!
--Karl Glazebrook, kgb (at) pha (dot) jhu (dot) edu
Measuring European air pollution for the U.N.
The UN-ECE's EMEP program monitors long-range air pollutants such as acid rain and ozone, over the whole of Europe. The results are used to create international laws and accords such as the Gothenborg protocol. Our FORTRAN code models pollutant motion and chemistry throughout Europe, producing about 50 GB of netcdf data for each year studied.
Comparing these data with other data sources, extracting country-based data from the physical grid, and preparing statistics required a powerful, memory-efficient, flexible data language. PDL gave us exactly what we needed. We even used it to generate automated reports for each of 50 countries.
--Heiko Klein, Heiko.Klein (at) gmx (dot) net
Forest fire detection
At the Remote Sensing Laboratory of the University of Valladolid, we combine data from several satellites to produce real-time images of Spain and surroundings as a public service. Among other services, we can detect forest fires from satellite images within a few minutes, in time to alert regional authorities. Our pipeline includes 20 steps, some compiled in C and FORTRAN and others in Perl and Perl/PDL. Our results are published on the web and emailed to Televisión Española where they are made available via the national teletext service.
At right is an image of spain with several potential wildfires marked in red. Their locations are listed at far right. The various aqua-colored outlines are regions that were photographed by different satellites.
-- Joaquin Ferrero, explorer (at) latuv (dot) uva (dot) es
Visualization of my brain
perlfMRI is a very simple tool I created for viewing functional MRI data, written entirely in perl with PDL. What's nice about it is that it lets you simultaneously view three planes through the brain and to rotate them in 3d (see below). These images are showing my own brain, overlain with activations in my auditory cortex while I listened to music. They were created under MacOS X.
--Jonas Kaplan ( jonask (at) ucla (dot) edu )