Photogrammetry and non-destructive determination of properties of Antarctic meteorites

Tom Harvey writes:

Throughout ejection from their parent body, transportation through space and subsequent arrival on Earth, meteorites undergo extreme pressures and temperatures. Some meteorites, despite having survived entry to Earth, can be extremely fragile. Some of the analytical techniques that provide a wealth of information about meteorite geochemistry and physical properties can be damaging to a meteorite sample, either exposing it to some forms of contamination, or destroying it entirely. Given the opportunity to analyse relatively pristine samples from Antarctica, we opted to develop a procedure for understanding some of the fundamental properties of the returned meteorites that was entirely non-destructive.

Density (the mass per unit volume of a material) is a useful property to understand because it can allow us to make inferences about parent body composition, in addition to being an important metric in understanding thermal evolution models, as well as the survivability of materials in impacts, and during atmospheric entry to Earth. Non-destructive, high-precision measurement of the mass of a meteorite is straightforward to undertake using lab scales (i.e., weighing balances).

However, the determination of sample volume is not so straightforward. Traditional “Archimedean” methods for the determination of volume involve measurement of the displacement of a well-characterised medium (for example water, ceramic beads or inert gases) when the sample is emplaced into (see this previous study Consolmagno et al. 2008). Depending on the medium that the meteorite might come into contact with, it is possible that the sample might be chemically altered or mechanically damaged. Volume determination has also been performed by laser scanning of meteorite samples, to produce a three-dimensional (3-D) computer model of the sample surface topography (for example, see McCausland et al. 2011).  A potential barrier to performing this sort of study is the requirement for expensive and highly specialised laser and camera technology.

An alternative to these methods that we have chosen to develop is the production of 3-D computer models of the samples returned by the Lost Meteorites of Antarctica project. This method involves the use of a suite of photographs of an object to computationally generate a three-dimensional colour model of that object. The method is non-contaminating, scalable, and the only necessary hardware was a DSLR camera, some small portable lights, and a computer with the appropriate software installed.  Furthermore, the production of an accurate computer model of the samples’ surface topography and colour makes an excellent curatorial record which can be used in lab decision making, and to understand the orientation relationship between sub-splits in the future.

Tom Harvey in the curation lab imaging one of the meteorite samples. Photo: KJoy

First, the meteorites were placed into a controlled light environment. This meant that we could control the amount of reflection and shadow on the sample surface, which is important for producing a good model. By using the appropriate sterile tools and portable lights, this method did not expose the samples to any alteration or contamination. Photographs were taken using a high-resolution camera at 5º rotational intervals around the sample (and in multiple orientations to capture) to ensure that we had captured all the details of the sample surface.

These photographs were then loaded into Agisoft Metashape, a professional software for photogrammetry applications, which were used to produce the models. This process is broken into several key steps which involve a pixel matching algorithm, and alignment of the various orientations of the sample to generate a shape file for the sample, followed by the production of a mosaic derived from the sample photographs to map over the shape file to bring it to life in colour.

In order to determine the volume of these models, the shape files were imported into a computer-aided design (CAD) software, 3DS Max, and scaled according to known external sample dimensions, measured in the lab with a Vernier calliper. In order to understand how well this method determined the true volume of the sample, we produced models of wooden cuboid blocks of known size and found that the computed volumes of these blocks were within 2% of their true, measured volume.

Final model of Antarctic Meteorite OUT 18010, images of which are shown in the above slideshow. Video: Tom Harvey

Using the data from this method, we will be able to compare our computed volumes with measurements derived from other methods such as computed tomography scans of the sample. The photogrammetry derived volume measurements and the sample masses measured in the lab allowed us to compute a density value for each meteorite in the study. Once all of the meteorites are formally classified, we will also be able to compare the densities derived from our photogrammetry-based method, with literature density values for meteorites of the same type as the ones recovered by the Lost Meteorites of Antarctica team to assess the usefulness of the method.

One of the potential drawbacks of our method is that it can be fairly time consuming, taking several days to carefully photograph, and computationally process a meteorite from start to finish. Furthermore, the method may be challenging to undertake on highly reflective metallic samples – the software relies on identifying matching features in different pictures, and so a sample that predominantly reflects light might cause issues.

However, it is possible to make high fidelity models of meteorites of a range of a size, shape and colours. These can be useful curational tools and should also provide a dataset of information about the samples that would otherwise be difficult to attain whilst maintaining the pristine nature of the samples.

This study was the topic of a 2021 LPSC abstract which you can find here. You can find the various models we’ve made dotted about the individual meteorite information pages from the first season, or you can find a selection of them below!


Testing the AnMetMet

Tom Harvey | 07 Dec 2018

One of the important parts of collecting meteorites is detailed record collection which allows for efficient curation and identification when the samples are returned from Antarctica. In addition to where a sample was found, we need to know what it is (and be sure that it is a meteorite after all!). As mentioned in a previous post, in addition to her familiarity with collecting meteorites in the field, one of the ways that Katie might be supplementing her preliminary in-field examination of the samples is by using a CEREGE-built portable combined magnetic susceptibility – conductivity field probe. Magnetic susceptibility is a measurement of the degree to which a material will be magnetised (induced magnetisation) when a given magnetic field is applied. Conductivity is an electrical property which allows a material to conduct electricity.

Photo of the probe in use with the set of test meteorites [Image: T Harvey].

Following an initial blank air test, the sample is placed close to the probe, reacting with the weak magnetic field generated by the probe. The extent of the reaction is correlated to magnetic susceptibility, and also affected by electrical conductivity. Both properties are then quantified using correction based on sample volume (Gattacceca et al., 2004).

The predominant component of meteorites which can be magnetised is the meteoritic iron (i.e., iron as Fe rather than oxidised phase like FeO or Fe2O3). The meteoritic native iron content of meteorites varies quite reliably with the classification of its type (stony, stony-iron, iron), which depends on its composition, thus linking to its original process of formation. Meteoritic iron is also electrically conductive, whilst terrestrial rocks usually are not. As such, the AnMetMet is able to tell the difference between metal-bearing samples and iron oxide (e.g. Magnetite) bearing samples, a separation that cannot be achieved using a magnet which is sensitive to magnetic susceptibility alone.

Because of this, performing a magnetic susceptibility-conductivity test gives a good estimate of its classification – and can be used to confirm that the rock is indeed a meteorite rather than a terrestrial magnetic rock or mineral. A sample must be above a certain volume for accurate readings, but if they are relatively homogeneous, even large samples (larger than the probe) can be measured.

The magnetic field generated by the probe is weak, similar in intensity to the natural Earth field, so that the natural remanent magnetisation (NRM) of the sample is not disturbed. This is crucial because the NRM of meteorites is of great scientific value as it bears the record of ancient magnetic fields that may have been present in the early solar system, generated within the solar nebula or by dynamo phenomena within planetesimals. For that reason, the use of magnets to test the meteoritic nature of a rock should be avoided at all costs because it both inefficient (non quantitative) and destructive for the magnetic memory of the rock.

To ensure that we understand the numerical output of the probe in the field, we are testing it on some test meteorites we already have here in Manchester (these are the same samples we have been testing our metal detector panels on). Amongst the test subjects are pieces from NWA-869 – an ordinary L chondrite, Gao Guenie – an ordinary H chondrite, Campo del Cielo – an IAB iron meteorite and a few fragments from the ordinary LL chondrite Chelyabinsk fall (which some might remember from the dash-cam videos in 2013).

Test meteorite samples: CH2 is a piece of Chelyabinsk, B7 is a piece of NWA-869, C7 is a piece of Gao Guenie and D7 is a piece of Campo del Cielo. D7 is clearly much more metal rich than the previous 3, but the metal content of the other samples vary too [Images: A R D Smedley].

The aim is to piece together a reliable framework of readings so that we can be sure of how potential finds fit into the meteorite classification scheme. In addition to this, we will be testing the probe at low temperatures to see whether the readings will vary from room-temperature measurements when they are faced with the significant chilliness of Antarctic weather.

If all goes according to plan, we’ll have a useful database to complement Katie’s in-field measurements, which will be a good aid for future sample curation planning and analysis. Big thanks to Jérôme Gattacceca for making sure this explanation is accurate!