Abstract: Rock is the basic material of Earth. Microscopic investigation is instrumental method in earth sciences. However, there are currently no unified standards for rock micrograph acquisition in geology, preventing the establishment of comprehensive databases for rock microscopic images. Therefore, we launched this special issue as a springboard to systematically integrate the photomicrograph data of sedimentary rocks while facilitating a unified standard for data collection, entry, and storage. In this paper, we explored standards for dataset development based on both simple and derived information from micrographic observations of thin sections of three major groups of sedimentary rocks, namely, limestones, sandstones, and mixed siliciclastic and carbonate rocks. The standards delineated herein serve as a framework for data collection and processing in the studies in this special issue and laters.
Keywords: micrograph; unified standard; sandstones; limestones; mixed siliciclastic and carbonate rock; database
Rocks are natural mineral assemblages exhibiting a stable shape. The crust and upper mantle of the Earth are mainly composed of rocks. Rocks are classified into three major types based on different dominant processes: igneous, sedimentary, and metamorphic rocks. The earliest descriptions of rocks can be found in the book Shanhaijing, which dates back to the pre-Qin period. The “fire and water controversy” pertaining to rocks among Hutton J (1726–1797), Lyell C (1797–1875) resulted in the birth of modern geology more than 200 years ago. Since Sorby’s study of the polarized light microscope in 1880 on the rock, the discipline of petrography, which is largely based on thin sections of rocks, has gained increasing popularity. Petrography has played an irreplaceable role in the study of geology and is still underway today, even though analysis and testing techniques have significantly evolved. Under a microscope, it is possible to observe the previously invisible phenomena of rock composition and interrelationships, greatly aiding humans to understand Earth’s laws and mysteries. Photographs of thin sections of rocks obtained using a polarizing microscope are called rock micrographs in this study.
For a long time, scientists have been unable to develop a unified standard for rock micrographs and so far, limited unified databases of rock micrographs are available. Scientists, research groups, or institutions only capture a small number of images according to their own requirements and objectives and publish them in academic papers or on the Internet as a part of scientific research results or display materials. For example, the British Geological Survey posted a very small number of microscopic images of different rock types on its official website. Another main reason for the lack of microscopic rock image databases is the limitation of digital image capturing and storage technologies in the past. With recent advancements in digital image technology, the large-scale capture and storage of rock micrographs has become possible and practical. As increasing digital microscopic rock images are generated, it is essential to unify the standards and processing procedures of these images to allow for future integration of microscopic rock image data.
Herein, the datasets of rock micrographs are established to facilitate easy sharing and use by humans and computers. The use of these datasets by humans is well understood—scientists need more datasets to conduct comparative studies and public science. The use of the datasets by computers refers to the fact that with the rapid development of image and artificial intelligence technology, microscopic-image-based studies have become possible. An important prerequisite for achieving this research model is the requirement of different rock types, a certain number of datasets, and uniform standard and information entry format for effective integration of these data.
The dataset developed in this study mainly contains sedimentary rock micrographs. The main content and discussion also include the acquisition and processing standards of microscopic sedimentary rock images (Figure 1). The construction of a dataset containing information of magmatic rocks, metamorphic rocks, and other rock types can follow the acquisition and processing standards proposed herein.
2.1 Rock collection and thin section production
By focusing on certain scientific research targets or production tasks, geologists have investigated thin sections, measured stratigraphic sections, and collected rock samples from well-exposed, fresh outcrops. To meet the requirements of thin section production and hand specimen observation, the size of the rock samples used herein is generally maintained to be not less than 3 cm × 6 cm × 9 cm. Based on the research objectives, strata thickness, lithological changes and sedimentary characteristics, the different types and numbers of rock samples are collected.
The thin section of the rocks is fabricated by professional companies or researchers followed the standard production process of rock thin sections. The rock samples are processed into 0.03-mm-thick thin sections to be directly observed under a polarized light microscope.
2.2 Acquiring polarized light micrographs of rock
First, an operator must select a representative field of view on the thin section to achieve at least one plane-polarized light micrograph and one cross-polarized light micrograph under the same view field at a certain magnification ratio. If the composition and texture of the rock are nonhomogeneous, additional micrographs would be captured in different view fields, if necessary. If there are interesting textures, structures, fossils, etc., the operator can supplement the micrographs and sequentially label them with numbers. The numbering standard is “thin section number” + “m” + “the serial number of the micrograph on different view fields” + “cross-polarized light symbol (+) or plane-polarized light symbol (−),” where “m” represents a micrograph. Considering the thin section sample numbered 16BK75 as an example, the first captured plane-polarized and cross-polarized light micrographs are termed 16BK75m1− and 16BK75m1+, respectively.
The magnification of the micrographs is performed to objectively express the characteristics of the thin sections by considering the identification of minerals or grains, textural information, etc. A scale bar is added to each micrograph and uniformly placed in the lower right corner (units in μm) (Figure 2). During the micrograph capture process, the automatic exposure and white balance functions are used to ensure that the color of the micrographs are the same what the unaided eyes see. The resolution of the micrographs is maintained uniform by using the highest resolution of the camera system. For example, the camera system of the Nikon polarizing microscope in the Optical Microscope Lab of Nanjing University School of Earth Sciences and Engineering has a resolution of 4908 × 3264 pixels. The micrographs are uniformly saved in the PNG or JPG format.
2.3 Basic input information of rock thin sections
The basic input information of the rock thin sections is shown in Table 1. It mainly includes the sample location information of the administrative division or global positioning system (GPS) coordinates, age and stratigraphic unit of the rock samples. Moreover, the basic input information includes the name and institution of the owner of the thin sections and related published literature or literature ID. This information will facilitate access to academic papers closely related to this dataset, contact thin section holders for future cooperation, etc.
|References||Location||Section Names||Latitudes||Longitudes||Group/Formation||Age||Thin Section|
|Zhang, S. et al., 2018, Palaeoclimatology, Palaeoecology 501, 92-110.||P.R. China||Xinjiang||Kizilsu Kirgiz Autonomous Prefecture, Akto County, Kiziltao town, and Qimigan village||Qimugen section b||38°25′56.26″||76°23′41.86″||Lower part of the Qimugen formation||Ypresian–Thanetian||Hu Xiumian and Zhang Shijie|
Note: the "References" section header is optional and the others are mandatory.
Additionally, the thin sections contain two types of information. One is micrograph information that can be observed repeatedly, such as the micrograph of the thin sections and the basic information of the rock thin sections. The other is the describable information such as the description, classification, and name of the thin sections. The reference standards and procedures for obtaining indirect information of the rock thin sections are shown below.
3.1 Rules for classifying the sedimentary rocks
One of the tasks related to the identification of the thin sections is classifying the type of the observed rock (Figure 3). The “rock type” in the appraisal tables of limestone (Table 2) and sandstone (Table 3) refers to the type of sedimentary rock, including conglomerate, sandstone, limestone, siltstone, and mudrock. The sedimentary rocks also include dolomite, evaporite, siliceous rock, and mixed siliciclastic and carbonate rock (Figure 3). This special issue mainly focuses on these rock types, namely, limestone and sandstone. A small amount of mixed siliciclastic and carbonate rock, siltstone, dolomite, mud shale, and evaporite is also observed.
Limestone is classified according to the classification proposed by Dunham (1962)  and Embry and Klovan (1971) , which is mainly based on the content of grains and cement in the thin sections, grain types, supporting texture, and in situ organisms (Figure 4A).
Figure 4 The classification of limestone (A), sandstone (B), and mixed siliciclastic and carbonate rock (C)
Classification of limestones is based the classification proposed by on Dunham (1962)[ 1] and Embry and Klovan (1971)[ 2], the classification of sandstones is based on the simplified classification by Garzanti (2016)[ 3], and the classification of mixed siliciclastic and carbonate rock is based on the classification proposed by Mount (1985)[ 5]. For the symbol of QFL, refer to Table 4.
The four-component classification principles are adopted to classify sand and sandstone. First, the simplified classification proposed by Garzanti (2016) [3-4] (Figure 4B) is used to directly obtain the basic names of six types of sandstones, namely, litho-quartzose, feldspatho-quartzose, litho-feldspathic, quartzo-feldspathic, quartzo-lithic, and feldspatho-lithic sandstones based on the relative content of quartz, feldspar, and lithics. If the content of quartz, feldspar, or lithics is particularly enriched, then whether the content of the other two components is less than 10% or not is determined and the content of quartzose, feldspathic, or lithic sandstone is further determined. Second, depending on whether the content of the matrix is greater than 15%, it is classified into greywacke or sandstone, such as quartzose sandstone or quartzo-lithic greywacke . Essentially, it is not necessary to estimate the specific contents of quartz, feldspar, or lithics but to determine their relative proportions. The rule of classifying sandstones is simple to use and significantly reduces errors caused by human unaided eye estimation.
For mixed siliciclastic and carbonate rocks, the classification proposed by Mount (1985)  is used, which is based on the relative content of siliciclastic grains, carbonate grains, terrigenous mud, and carbonate mud (Figure 4C).
3.2 Standards for data collection of limestone under a microscope
The identification table of limestones describes the type of limestone subdivision (Table 2). Based on the classification proposed by Dunham (1962)  and Embry and Klovan (1971) , limestones can be divided into allochthonous limestones (with grain structures), autochthonous limestones (with biological structures)), and crystalline limestones (with crystallized structures) (Table 2). The operator can select the header based on the type of limestone. For example, for the allochthonous limestones with granular structures, only the header of the granular structures needs to be filled in (Table 2); for the autochthonous limestones with biological structures, only the header of the biological structures needs to be filled in. The header of the crystalline limestones with recrystallized structures should be filled in depending on whether the structure is preserved or not. For recrystallized structures without structural residue, only the header of the recrystallized structures should be filled in.
|Thin section number||Rock type||Name||Granular or original texture is recognizable|
|Carbonate grains||Groundmass types||Content of micrite||Supported texture||Supported texture with grain sizes of >2 mm|
|Carbonate grain types||Content||Total content||Size||Sorting||Skeletal grain preservation degree||Grain content with >2-mm size|
|15QM01||Mudstone||Crinoid||5%||5%||1–2 mm||Subhedral||Micrite||94%||Mud supported|
|Planktonic foraminifers||<1%||<0.1 mm||Euhedral|
|Granular texture or original texture is recognizable||Crystalline texture||Biological structure or original texture is recognizable||Supplementary description|
Supported texture with grain sizes of >2 mm
|Crystal size||Crystal texture||Crystal shape||Euhedral degree||Biological structure||Biological buildup type|
|Other grains||Cement fabric||Sedimentary texture||Sedimentary structure||Diagenesis||Others|
Note: the bold headers are mandatory and others are optional.
In limestones, the estimation of the grain content with a particle size of >2 mm is requisite for naming. Herein, the method for calculating the particle content is as follows: the calculations of the grain content with a particle size of >2 mm and total grain content are independent of each other. The total grain content is the sum of all types of grain listed in the sample; the grain content with a particle size of >2 mm refers to the sum of the contents with particle sizes greater than 2 mm. For example, limestones contain bivalve debris and foraminifers. The total content of bivalve debris is 20%, and the content of bivalve debris with a particle size of >2 mm is 5%. The total content of foraminifers is 25%, and the content of foraminifers with a particle size of >2 mm is 10%. Then, the total content of grains is 45% and the content of grains with particle sizes of >2 mm is 15%.
The classification scheme of the skeletal grain preservation degree in the limestone identification table (Table 2) adopts the classification standard proposed by Yu (1989)  (Figure 5). It is divided into euhedral, subhedral, sand-to-gravel-sized anhedral, and silt-sized anhedral. The type of groundmass includes micrite and sprite. The micrite size is defined by Embry and Klovan (1971) , corresponding to calcite crystals with a diameter less than 0.03 mm. The crystal size is based on the limestone classification standard proposed by Zeng Yunfu (1986) . The crystal texture and shape are based on the standard proposed by Fridedman (1965) . The biological structure is based on the limestone classification proposed by Embry and Klovan (1971)  (Figure 4A). The biological buildup type is based on the limestone classification standard proposed by Zeng Yunfu (1986)  (Table 2). The particle sorting is based on the classification proposed by Jerram (2001) , in which a visual comparator is obtained that shows the sortability in a slice, including very well sorted, well sorted, moderately well sorted, moderately sorted, and poorly sorted (Figure 6).
3.3 Standards for data collection of sandstone under a microscope
The description of clastic rocks mainly focuses on detrital components (composition type, content, etc.), interstitial materials (matrix, cement, etc.), special sedimentary texture, diagenesis, etc. (Table 3). Among them, the identification of depositional textural information (such as size, roundness and sorting) based on visual observations, which can easily vary among people, is temporarily excluded. This information can be assessed using the micrograph according to unified standards by adopting automatic image recognition technology in the future. The column of lithic fragment type and content in Table 3 represents the types of lithic fragment and their proportion in total lithics, and each type of lithic fragment can be represented using its unique code (Table 4). The matrix content refers to the percentage of fine detritals with particle sizes of less than 0.03 mm in the view field of the microscopic image. This information is used only to determine whether the matrix content exceeds 15% and select one of the following three drop-down options: “≥15%,” “<15%,” and “not applicable.” The cement type is optional and includes siliceous, calcareous, ferriferous, clayey, and other cement types commonly found in clastic rocks. Diagenesis is used to provide possible observations of compaction, pressure solution, cementation, metasomatism, recrystallization, dissolution, and other sedimentary diagenesis phenomena. To facilitate the supplementary description of non-sandstone micrographs, Table 3 presents a specially designed column termed “Others” for users to fill in.
|Thin section number||Rock type||Classification (Garzanti, 2016)||Particle description||Matrix content||Supplemental description|
|Particle content more than 10%||QFcontent||Lithic fragment type and content||Cement type||Diagenesis||Others|
|15TY78||Sandstone||Litho-quartzose sandstone||Q, L, and F||Q ＞ L ＞ F||Lv and Lc||＜15%||Calcareous||Bioclastic and Lc are intraclast|
Note: the bold headers are mandatory and others are optional.
|Qm||Monocrystalline quartz||Single quartz|
|Qp||Polycrystalline quartz||e.g., quartz vein|
|Q||Total quartz||= Qm + Qp|
|Pl||Plagioclase||Size≥ 0.063 mm|
|Kf||K-feldspar||Size≥ 0.063 mm|
|F||Feldspars||= Pl + Kf|
|Lv||Volcanic rock fragments||Size≥ 0.063 mm|
|Lu||Ultrabasic rock fragments||Size≥ 0.063mm|
|Ld||Detrital lithic fragments||Size≥ 0.063mm|
|Lc||Carbonate lithic fragments||Size≥ 0.063mm|
|Ls||Sedimentary lithic fragments||= Lc + Ld + Chert|
|L||Lithic fragments||= Ls + Lv + Lm|
|Lt||Total lithic fragments||= Ls + Lv + Lm + Qp|
|Mat||Matrix||Fine fragments with particle sizes of <0.03 mm|
|Acc||Accessory minerals||<1% in volume|
|Cc||Calcite||Brilliant calcite cement|
3.4 Standards for data collection of mixed siliciclastic and carbonate rocks under a microscope
Mixed siliciclastic and carbonate rocks considered in this study are those found in mixed siliciclastic and carbonate rocks. Mixed siliciclastic and carbonate rock is either a terrigenous clastic rock that contains more than 10% carbonate or more than 10% terrigenous clastic components in carbonates. Therefore, it is necessary to objectively and simultaneously describe the composition characteristics of carbonates and terrigenous clastics in detail in the standard table of the micrograph dataset. In this standard formulation, the use of the carbonate rock identification sheet to describe and record mixed siliciclastic and carbonate rocks is recommended owing to the following factors. (1) There are detailed unified descriptions and record formats of carbonate grains and cement in the carbonate rock identification sheet, which can meet the description of carbonates in mixed siliciclastic and carbonate rocks; (2) The supplementary description part in the carbonate rock identification sheet provides a special table, which can meet the general description of terrigenous clastics, such as cement, sedimentary textures and structures, diagenesis, and other characteristics; (3) Owing to the complex types of particle in carbonate rocks, the use of the carbonate rock identification sheet is more convenient and comprehensive than the use of the sandstone identification sheet to describe the carbonate rock information of mixed siliciclastic and carbonate rocks; (4) Although mixed siliciclastic and carbonate rocks are common in nature and are found in geological history, they are relatively less likely to appear than sandstones or limestones. If another standard of the identification sheet of mixed siliciclastic and carbonate rocks for less commonly found mixed siliciclastic and carbonate rocks is established, the operator will face additional difficulties when using datasets.
It is necessary to first indicate that the mixed siliciclastic and carbonate rocks are described in the Table 2. Then, according to the classification and naming scheme of the mixed siliciclastic and carbonate rocks, detailed name is defined by rules (Figure 4C). The contents of siliceous detrital grain, noncarbonate mud, carbonate grain, and carbonate mud must be recorded for naming the mixed siliciclastic and carbonate rocks. Therefore, the particles and cement type must be recorded in detail with reference to the description of the limestones. For the siliceous detrital grain and noncarbonate mud, it is necessary to record the content, cement type, textures and structures, diagenesis, and other characteristics (such as lithic fragment type) of terrigenous clastic rocks as special particles in the supplementary description. The identification content of carbonate rocks and terrigenous clastic rocks has been discussed in subsections 2.2 and 2.3.
3.5 Description of data compatibility
The rock microscopic image dataset in this issue mainly comprises microscopic images, field geological information, rock thin section information, measured stratigraphic sections, and other related information. The complete information can comprehensively present the basic information of all aspects of rock micrographs. This information can also effectively assist in achieving correlation or compatibility with other types of datasets.
The format of the high-resolution rock micrographs collected in this issue can easily be converted to other formats, such as picture formats, for computer reading, convenient storage, science education, and other purposes.
Furthermore, the dataset contains multiple data interfaces, which can be correlated well with different fields of research or applications. The popularity of this dataset will be aided by the compatibility of correlations. For example, the geographic locations or GPS in the dataset can be effectively used for tectonic studies or linked to social applications. Information such as stratigraphic units or epochs can be easily used for basic geological studies.
Although these standards are based on sedimentary rock types, mainly sandstones and limestones, they can help in the future development of a uniform format and facilitate data sharing of microscopic images of other types of rocks and minerals.
The information collected in this issue is first-hand data. The information tables contain descriptions of different fields. Based on the collected basic information, it is simple to effectively classify or filter rock micrographic images to meet our needs.
These shared data, including geographic location, thin section owners, and other contents, can facilitate further collaborative research. This will promote geological cross-collaborative research and expand the study field.
We thank all members of the team for their constructive discussions and comments on the construction standards for microscopic image dataset. This paper was cofunded by the NSFC PROJECTS (41525007; 42050102) and the National Key Research and Development Program of China (2018YFE0204201).
 DUNHAM R J. Classification of carbonate rocks according to depositional textures. AAPG Mem, 1962, 1: 108-121.
 EMBRY A F, KLOVAN J E. A late Devonian reef tract on northeastern Banks Island, NWT. Bull Can Petrol Geol, 1971, 19: 730-781.
 GARZANTI E. From static to dynamic provenance analysis-Sedimentary petrology upgraded. Sedimentary Geology, 2016, 336: 3-13.
 HU X, LAI W. Sedimentary Petrology: Chapter 6: Terrestrial clastic rocks. Beijing: Science Press, 2019: 118-158.
 MOUNT J. Mixed siliciclastic and carbonate sediments: a proposed first order textural and compositional classification. Sedimentology, 1985, 32: 435-442.
 FRIEDMAN G M. Terminology of crystallization textures and fabrics in sedimentary rocks. Journal of Sedimentary Research, 1965, 35(3): 643-655.
 JERRAM D A. Visual comparators for degree of grain-size sorting in two and three-dimensions. Computers and Geosciences, 2001, 4: 485-492.
How to cite this article
HU XM, LAI W, XU YW, et al. Standards for digital micrographs of the sedimentary rocks. China Scientific Data, 2020, 5(3). (2020-09-15). DOI: 10.11922/csdata.2020.0008.zh.