An Ordovician to Silurian graptolite specimen image dataset for global correlation and shale gas exploration

A unique high-resolution image dataset consists of key graptolite species used for dating rocks, global correlation, and “gold caliper” for locating shale gas favourable exploration beds (FEBs) in China.

All images were taken from 1,550 carefully curated graptolite specimens, taxonomically belong to 113 graptolite species or subspecies. These specimens were collected from 154 representative geological sections of the Ordovician to Silurian sediments of China and published in 1958-2020. All specimens are housed at the Nanjing Institute of Geology and Palaeontology (NIGP), Chinese Academy of Sciences (CAS). Detailed scientific information of every piece of fossil specimen is given in the attached spreadsheet file.

My working group spent over two years to complete photographing every specimen using a single-lens reflex camera Nikon D800E with Nikkor 60 mm macro-lens and Leica M125 and M205C microscopes equipped with Leica cameras. Every image is well focused and better shows the morphology of graptolite bodies.

In total, we took 40,597 images, including 20,644 camera photos (each with a resolution of 4,912 × 7,360) and 19,953 microscope photos (each with a resolution of 2,720 × 2,048). Photos of low contrast or bad focus were removed from the whole collection. We only kept and selected the photos that show the visual morphology of every specimen and the diagnostic character of each graptolite species that the specimens represent. We selected one image for each specimen as the present final dataset, uploaded to and stored in our cloud server.

We incorporated revision suggestions from distinguished palaeontologists to generate the ground-truth labels, providing a taxonomical authority of the dataset. The dataset potentially contributes to a range of scientific activities and provides 1) easy access to high-resolution images of 2951 specimens of 113 graptolite species for teaching and training in palaeontology and geologic survey; 2) Global bio-stratigraphic correlation using graptolites, especially with those bio-zone species; 3) A standard fossil specimen image dataset used in shale gas industry to improve exploration efficiency, and 4) The potential aid of developing image-based automated classification model.

Every specimen has two photos, one is original, another shows specimen with a scale bar. Occasionally in some large image the scale bar is embedded and beside the fossil specimen. Example: The file name: ‘9721Cardiograptus_amplus_S.jpg’, ‘9721’ is the specimens number, ‘Cardiograptus_amplus’ means species name is ‘Cardiograptus amplus’, with ‘_S’ means it is a photo with scale bar. In all scale bar, the minimum unit is millimeter.

Author and contact:

Hong-He Xu

Nanjing Institute of Geology and Palaeontology, Chinese Academy of Sciences

39 East Beijing Road, Nanjing, 210008

China

E-mail: hhxu@nigpas.ac.cn