RockSL: an integrated rock spectral library for better global shared services

Figures & data

Table 1. The characteristics of shared spectral libraries integrated in RockSL.

Spectral library	Sample Types			Wavelength range^b			Shared format	Data resource
	Rock/mineral	Other materials^a	Planetary materials	VNIR	MIR	FIR
USGS	✔	✔	✔	✔	✔	✔	ASCII	https://www.usgs.gov/labs/spec-lab/capabilities/spectral-library
JHU	✔	✔	✔	✔	✔		ASCII	http://speclib.jpl.nasa.gov
JPL	✔			✔			ASCII	http://speclib.jpl.nasa.gov
PDS	✔		✔	✔	✔		ASCII	https://speclib.rsl.wustl.edu/search.aspx
ASU	✔		✔		✔	✔	ASCII /HDF	https://speclib.asu.edu/
MISA	✔			✔	✔		Image	http://www.organchem.csdb.cn/scdb/

^aOther materials include soil, vegetable, snow, ice, water, artificial materials.

^bVNIR (0.3–2.5 μm), MIR (3–25 μm), FIR (25 –1000 μm).

Table 2. The metadata variables of data model in RockSL.

Group	Variable	Explanation	Metadata type
Common basic information	Mineral/rock code	Code of mineral/rock based on classification scheme	Categorical
	Sample code	Code of each specimen	Categorical
	Chinese name	Name of specimen in Chinese	Categorical
	English name	Name of specimen in English	Categorical
	Type code	Code of type of specimen (e.g. halides)	Categorical
	Locality	Spatial sampling position (longitude/latitude/altitude)	Quantitative
	Chemistry	Chemical composition	Alphanumeric
	Ideal chemistry	Chemical formula	Alphanumeric
	Particle size	Particle size of measured specimen	Quantitative
	Pictures	Image depicting the specimen/sampling environment	Pictorial
Signal properties	Spectral coordinate	Spectral data	Quantitative
	Wavelength range	Wavelength range	Quantitative
	Wavelength interval	Wavelength interval	Quantitative
	Data type	Data type (reflectance, radiance …)	Categorical
Shared data details	Database name	Name of shared database	Categorical
	All Information	Original information downloaded from libraries	Alphanumeric
	URL	URL link of target library	Alphanumeric
	Description	Descriptive information of target library	Alphanumeric
Measurement information	Sampling time	Date and time of the sampling in UTC	Quantitative
	Measurement type	Directional-hemispherical, bidirectional, hemispherical-directional reflectance	Categorical
	Instrument/spectrometer	Specific instrument	Categorical
	Measurement temperature	Air temperature in degrees	Quantitative
	Measurement pressure	Air pressure in hPa	Quantitative
	Measured environment	Field or laboratory	Categorical
	Contributor	Name of contributor	Alphanumeric
	Measured distance	Distance of the sensor to the target	Quantitative
	Measured angle	Zenith angle between sensor and specimen	Quantitative
	White reference	White reference plane used	Categorical
	Number of averaged spectra	Number of spectra averaged internally by the instrument	Quantitative
File information	File name	File name of the spectral file	Alphanumeric
	File format	File format of the spectral file	Alphanumeric
	Image ID	Code of curve image through vectorization of extraction of digital coordinate	Categorical
Data evaluation	Data level	Data quality rating based on the parameter integrity (characterized degree)	Categorical
	Data feature	The type of spectral data based on a customized classifier	Quantitative

Figure 1. Data structure of the designed database schema. The relational table of mineral/rock code and mineral/rock type records the classification information. The table of mineral/rock attribute records the physical and chemical properties, while spectral data information records spectral data and related metadata. The table of instrument information records the spectrometer information. The numbers (i.e. 1 and m) represent the number of records in each table, indicating that one record in the table (e.g. code of mineral/rock type) corresponds to multiple records in another table (e.g. code of mineral/rock).

Figure 2. The processing workflow of RockSL. The sequence numbers on the left represents the corresponding data table in the spectral library, which was operated in corresponding steps.

Figure 3. The original image of the spectral curve (left panel) and its corresponding vectorized digital curve (right panel). The original images of muscovite and quartz as examples were downloaded from USGS and MISA spectral libraries, respectively (Kokaly et al., 2017; Peng & Liu, 1982).

Figure 4. The matching result of quartz sample. (a) The matching module with three reference spectrums (no. 3289, no. 3290 and no. 4440) of quartz sample from PDS Geoscience spectral library and ASU spectral library. (b) Comparison of unknown spectral curve and reference spectral curves of quartz.

Table 3. Matching results of spectral data of quartz specimen and reference spectra in RockSL.

Reference spectrum	Sample category	Total score	BC			SAM			SID			SCF
			Default	FD	CR	Default	FD	CR	Default	FD	CR	Default	FD	CR
No. 4440	Quartz (ASU)	8.015	0.891	1	0.925	0.987	0.354	0.044	0.976	0.988	0.473	0.987	0.354	0.036
No. 3290	Quartz (PDS)	7.951	0.982	1	0.919	0.98	0.431	0.043	0.963	0.728	0.465	0.975	0.431	0.034
No. 3289	Quartz (PDS)	7.379	0.807	1	0.839	0.961	0.334	0.035	0.913	0.798	0.381	0.957	0.334	0.02

Notes: Default (without preprocessing), FD (first-order differentiation), CR (continuum removal).

Figure 5. Visualization examples of a subspace projection in a 3D metadata cube: abstract model of establishing characteristic dataset of RockSL. The article elaborated several datasets of particle size, wavelength range and mineral component.

Table 4. The selection of metadata to establish the characteristic dataset.

Factor type	Variable	Explanation of dataset
Chemical composition	Mixture	Same mineral mixed with different materials
	Mineral type	Different samples from the same mineral type
	Content	Samples with different mineral content
Observation condition	Wavelength range	Spectral data from different wavelength range
	Observe angle	Spectral data observed from different angle
	Observe distance	Spectral data measured by different distance
	Measurement temperature	Spectral data measured in different environment condition
Surface morphology	Particle size	Same sample with different particle size

Figure 6. The reflectance spectra of hematite with different particle sizes. Spectral data from NEU measured in laboratory (Wang et al., 2018).

Figure 7. The full-band spectral data of calcite mineral from diverse libraries (i.e. USGS, PDS, JHU, ASU and MISA). The curve (ranges from 1.6–200 µm/6250-50 cm⁻¹) of USGS is displayed on the horizontal axis of wavenumber to highlight the characteristics of the far-infrared bands, while other curves are displayed on the horizontal axis of wavelength.

Figure 8. The spectra of calcite mineral with different mixture minerals measured by Analytical spectral devices (ASD) and Beckman instrument (downloaded from USGS Spectral library). Spectra are offset for clarity.

Figure 9. The data check result of hematite samples based on boxplot algorithm.

Figure 10. The data check result of multiple curves of monzonite sample based on internal conformity.

Kokaly, R. F., Clark, R. N., Swayze, G. A., Livo, K. E., Hoefen, T. M., Pearson, N. C., … Klein, A. J. (2017). USGS spectral library version 7: U.S. geological survey data series 1035. 61p, March, 2020. https://pubs.usgs.gov/ds/1035/ds1035.pdf

 Google Scholar

Peng, W., & Liu, G. (1982). In Zhou, M. S (Eds.).Infrared spectra of minerals (pp. 68–490). Beijing: Science Press.

 Google Scholar

Wang, D., Liu, S. J., Mao, Y. C., Wang, Y., & Tian-Zi, L. I. (2018). A method based on thermal infrared spectrum for analysis of sio_2 content in anshan-type iron. Spectroscopy and Spectral Analysis, 38(7), 2101–2106.

 Web of Science ®Google Scholar

Data availability statement

The integrated spectral library (RockSL) described in this article is openly available on GitHub at https://github.com/CSU-PCP-XBS/spectral-dataset-RockSL.