Desert Fireball Network

The Desert Fireball Network (DFN) is a network of cameras in Australia. It is designed to track meteoroids entering the atmosphere, and recover meteorites. It currently operates 50 autonomous cameras, spread across Western Australia and South Australia: Nullarbor plain, WA wheatbelt, and South Australian desert. The locations of the stations were chosen to facilitate meteorite searching.

Desert Fireball Network

Abbreviation	DFN
Type	Camera network
Purpose	Record meteorite falls
Headquarters	Perth
Region served	Australia
Affiliations	Curtin University
Website	http://fireballsinthesky.com.au

Science of Fireball Tracking

Meteorites found

Meteorite name	Fall observation date	Country	State, province, or region	Classification	Instrumentally observed - orbital data	Meteoritical Bulletin(s), other references
Bunburra Rockhole	July 21, 2007	Australia	South Australia	Brecciated achondrite	Yes
Mason Gully	April 13, 2010	Australia	Western Australia	H5	Yes
Murrili	November 27, 2015	Australia	South Australia	H5	Yes
Dingle Dell	October 31, 2016	Australia	Western Australia	L/LL5	Yes

Camera Hardware

Lambina DFN Station: a typical outback fireball observatory

The DFN observatories use consumer still photographic cameras (specifically DSLRs) with 8mm stereographic fish-eye lenses covering nearly the entire sky from each station. The cameras are controlled via an embedded Linux PC using gPhoto2 and images are archived to multiple hard disk drives for storage until the observatories are visited for maintenance (every 8–18 months depending on the storage capacity).

The observatories take one long exposure image every 30 seconds for the entire night. After capture, automated event detection searches the images for fireballs, and events are corroborated on the central server using images from multiple stations.

A GNSS syncronised time code is embedded in the long exposure images by the operation of a liquid crystal (LC) shutter to provide absolute timing data for fireball trajectories after triangulation with temporal precision better than one millisecond. Absolute timing is used for the calculation of meteoroid orbits and the relative timing also embedded by the timecode is required for trajectory analysis (specifically to calculate the mass from the deceleration of the meteoroid).

Internals of the latest iteration of the DFN observatory design (as of August 2017) displaying cameras, storage, power management circuit board and embedded PC.

Data processing pipeline

The rate of data acquisition requires an automated digital pipeline for data reduction. Neural network algorithms are used to detect fireball events. A wireless link to each Automated Fireball Observatory allows a cross-check for multi station confirmation and enables images to be remotely downloaded. Software has been created to facilitate the location of fireball trajectories in pixel coordinates. These are converted to celestial coordinates, to a minute of arc precision, by using a powerful astrometric calibration tool created to automatically identify surrounding stars, and use them as a referencing system. The different observation angles are triangulated using a modified least squares minimisation approach, which now includes weightings based on image quality to produce the full observed trajectory . A shutter system within the lens of each observatory encodes a unique non-repeating DeBrujin sequence into each fireball. This provides accurate, absolute timing information for the duration of the trajectory to 0.4 ms. Purpose written software uses entry parameters to determine orbits for each meteoroid. In order to determine if there will be a potential meteorite, the estimation of the changing meteoroid mass is modeled. Once ablation stops, the atmospheric winds strongly affect a meteoroid’s path to the ground. Data from the Global Forecasting System is used in an atmospheric wind model with a 0.008 degree resolution mesh uniquely created around the area of the fireball. A Monte Carlo dark flight simulation is performed to determine a likely search area for main mass and fragments.

Weather modeling

The dark flight trajectory of a meteoroid is significantly affected by the atmospheric winds, especially by the jet stream. As a result, the meteorite fall position can be shifted by up to several kilometers compared to a scenario with no winds.

The weather situation in the area around the end of the luminous flight is numerically modeled using the third generation of Weather Research and Forecasting (WRF) model with dynamic solver ARW (Advanced Research WRF). The weather model is typically initialized using global one-degree-resolution National Centers for Environmental Prediction (NCEP) Final analysis (FNL) Operational Model Global Tropospheric Analysis data. The model produces 3D matrix for given area and time, with horizontal resolution down to 1 km. From this 3D data, weather profiles are extracted; the components include wind speed, wind direction, pressure, temperature and relative humidity at heights ranging up to about 30 km altitude, in most cases fully covering the dark flight.

High volume data handling and archiving

The DFN produces 100s of terabytes of data per year, which mostly consists of high resolution all-sky images. With the proposed network expansion, this volume is going to increase. For the primary purpose of this network, meteorite recovery, only a small fraction of this data (images containing fireballs) is needed, and it is handled by the data processing pipeline (above). However, there are many other potential uses for the data is areas of Astronomy or Space Situational Awareness.

The full data volumes recorded by the cameras are too large to be transferred remotely. Removable hard drives are therefore collected during regular servicing of the DFN observatory sites, replaced with blank hard drives, and then transported to Perth to be archived in a data store at the Pawsey Supercomputing centre. The multi-petabyte data store allows searching of the dataset, using generic and project-custom metadata, and data sharing with other research groups.

Meteorite Searching

Meteorite fall predictions from a camera network typically produce a "fall line"—a straight or curved line on the ground typically a few km long—where it is believed the meteorite has fallen somewhere along the line, but its precise location is unknown. This is a result of the triangulation process, the effect of atmospheric winds during the fall, and knowledge of the apparent visible deceleration of the meteorite, but a lack of knowledge of its density, shape and precise mass.

Meteorite searching theory owes much to search and rescue theory, albeit somewhat simplified as the meteorite is not a moving target. Most of the falls observed by the DFN are in the remote outback, and so searching teams usually consist of 4-6 people, who camp on site for up to two weeks. This means that the searching strategy is focused on efficiency, rather than speed: meteorite recovery on the final day of the expedition is just as scientifically valuable as the first day, which contrasts to, for example, missing person search and rescue, where speed is of the essence. The practical searching techniques used by the DFN team are adapted to the predicted fall size and error ellipse:

• Searching on foot, gridding the area using GPS units to guide walkers, or using survey flags to mark areas, useful for smaller predicted masses, or a smaller error ellipse. This allows detailed coverage of the area with a higher confidence, but less area is searched per unit time.
• For larger areas, searching using quadbikes or ATVs. This is most applicable for larger predicted falls, or good clear area with good long distance visibility.
• Current research is focusing on the use of drones as a technique to improve efficiency.

See also

• Glossary of meteoritics
• European Fireball Network