Metadata are an important component in the cataloguing and analysis of in situ spectroscopy datasets because of their central role in identifying and quantifying the quality and reliability of spectral data and the products derived from them. This paper presents approaches to constructing a novel metadata standard for marine spectroscopy that serves to ensure a high level of reliability, integrity, and longevity for a spectroscopy dataset. Examined are the challenges presented by designing a standard that meets the unique requirements of in situ marine spectroscopy datasets, including the special case of measuring reflectance for underwater coral targets. Issues such as field measurement methods, instrument calibration, and data representativeness are investigated. The proposed metadata model incorporates expert panel recommendations that include metadata protocols critical to all campaigns, and those that are restricted to campaigns for specific marine environments. The implication of semantics and syntax for a robust and flexible metadata standard are also considered. Approaches towards an operational and logistically viable implementation of a schema are discussed. This paper also proposes a way forward for adapting and enhancing current geospatial metadata standards to the unique requirements of field spectroscopy. Data collection protocols, encompassing both field spectral measurement methods and the metadata associated with them vary widely across the breadth of scientific inquiry applied to in situ spectroscopy. Metadata is a central component to the reliability, integrity, and legacy of a spectroscopy dataset because it serves to mitigate systematic and random errors on recorded radiance, target discriminability and contrast (Duggin, 1985) and reduce system bias and variability (Pfitzner et al., 2006). On international and national scales, ad hoc data collection protocols are the norm as no formal standards exist within the remote sensing community for in situ metadata collection and rather arise from the expertise and knowledge of the scientists carrying out the campaign. Metadata recorded during a campaign may vary in format (hardcopy log sheets, excel forms, rudimentary database) and in volume (inclusive of documentation of all relevant campaign protocols to a minimum of metadata describing only the target being sampled). Metadata collection protocols diverge along the lines of the purpose of the campaign (calibration and validation, creation of a spectral library) and the target (tree crown, soil, seagrass, etc). Logistics, environment, instruments and target type all affect the design and implementation of a practical metadata standard. Here the special case of a metadata standard for a marine campaign for underwater coral reflectance is presented. Marine campaigns are unique from terrestrial campaigns in terms of the instrumentation, specialized requirements for in situ data collection and environmental variables. Targets can include seagrass, macro-algae, corals and sponges, spectral measurements may be taken above surface or below surface and opinions differ on how inclusive a metadataset must be to document environmental and target properties (Bhatti et al., 2009 and Dekker et al., 2010). Instrument housing is often necessary to permit submersion and in some instances the instrument must be specially adapted to the underwater light field. Spectral measurements are recorded in a potentially unsafe environment with often continuously variable viewing conditions (illumination, viewing geometry, turbidity, etc.). At the University of Queensland, a customized underwater spectrometer system was developed and tailored specifically to coral reef ecology, and the ecology and physiology of animal colour vision. The accompanying protocols
for recording metadata in situ are interdependent with the challenges of radiometric data collection underwater as they are designed to simultaneously ensure the requisite operator safety (Roelfsema et al., 2006).
To ensure a high quality and practical metadataset, a metadata standard for underwater coral reflectance should have the following properties: 1) the metadata fields are sufficient to comprehensively and explicitly document the activities that took place and quantify and qualify influencing factors to the spectral measurement 2) allow replication of the campaign if required 3) and be flexible and broad enough in the scope of data capture to permit interoperability with other datasets. Granularity (the degree of specificity of the variable being recorded), syntax of the fields, and their data format (numeric/text/timestamp) affects the potential for data export, mining, and sharing.
Presented here (Table 2.1) is a metadata standard for underwater coral reflectance spectroscopy. It is derived from input from an expert panel of marine remote sensing scientists at the ACEAS (Australian Centre for Ecological Analysis and Synthesis) Biooptical workshop held in Australia in 2012. While not inclusive of all metadata (instrument, calibration activities, reference standards, etc.) that should be recorded for an in situ campaign, it documents those metadata that describe field methods and variables unique to underwater coral reflectance measurements. The metatadaset is divided into four main categories: ‘Location and Environment Information’, ‘Illumination Information’, Viewing Geometry’, ‘Coral Target Properties’. A description and reasons for inclusion of each field is provided, as well an example of each. An optionality designation of either ‘Critical’ or ‘Useful’ has been assigned to each field. Assuming that campaign logistics are not always favourable to documenting all necessary metadata, a prioritization model for criticality can form the basis of a standard that is both practical and fits the purpose for which the data is being collected. Critical fields are those that ensure the integrity of the dataset and cannot be excluded; useful fields are those that increase the robustness of the dataset for purposes beyond which it was originally intended. The data type specifies the most suitable format (text/numeric/timestamp/binary/image) for a given metadata parameter. A ‘GML Object Type’ column is included to indicate those metaparameters that can be expressed as GML 3.3 (Geographic Markup Language) objects. GML 3.3 is an implementation of ISO 19107 (specifying conceptual schemas for geographic features) and is used here simply as an example of a vocabulary that could be used to implement the metaparameters as objects in a metadata schema. Reference to a standard vocabulary, such as that provided by GML, permits translating the standard into a schema with maximum interoperability.
The most populous category (23 fields) is ‘Location and Environment Information’. This is due to the high number of variables found within the marine environment that influence spectral measurements (water column properties, subsurface conditions, CDOM, etc.). There are commonalities with terrestrial campaigns (GPS coordinates, location description) but even in these cases special considerations must be made for the feasibility of recording these in situ. The ‘Illumination Information’ metadata category, while again sharing common fields with other non-marine campaigns, must make allowances for wave lensing and artificial light fields. The ‘Viewing Geometry’ category is identical to metadata requirements for most terrestrial campaigns except for documenting an operator’s position relative to the target when they must provide shading over the target with their body to compensate for the fluctuating light field. The ‘Coral Target Properties’ category, similar to ‘Location and Environment Information’, contains fields relevant to marine campaigns only and reflects the special requirements of documenting underwater coral reflectance measurements.
numeric text text text text 2m vis
Wave height and period (for radiance measures) Tide conditions
H or L Swell, wave height, long period waves
Wind speed
Wind direction Height of sensor from surface (if characterizing water column properties) Depth of sensor from surface (if profiling water column) Natural canopy
structure Suspended sediment concentration (for water column studies) Chlorophyll concentration
Secchi disk transparency/turbidity measure CDOM spectral slope CDOM concentration Detritus concentration Phytoplankton species/classes
Useful Critical/Useful Critical/Useful
Critical Critical Useful Critical Useful Useful Critical Critical Critical 1 m 5 kn Ssw photo filename #mgl -1 #mgl -1 M (?) -S value
text GML OBJECT TYPE gml:stringOrNull gml:CodeType Optical measure of ambient conditions
(direct, diffuse) Source of illumination (e.g. sun, lamp) Bulb intensity
Light spectrum Single beam/multi beam Beam coverage (as a degree measure) Time interval for weather station data logging Optical thickness of atmosphere Visibility estimate
Cloud cover % Cloud cover model Cloud cover threshold for this project
Wave lensing Natural canopy shading Artificial light canopy effect text text text boolean numeric
text
numeric gml:degrees Critical numeric numeric numeric numeric numeric text text text gml:doubleOrNull gml:Quantity gml:degrees gml:degrees gml:degrees gml:stringOrNull
Species or name
Coral species
Critical Critical
Name code Coral algae
etc. Coral Target Properties Metadata (continued) 30 cm text text image text image text gml:CodeType gml:Quantity gml:stringOrNull gml:stringOrNull gml:stringOrNull 5% 25˚ numeric
gml:Quantity
A viable and practical metadata standard for underwater coral reflectance measurements must provide flexibility for data sharing
in a common exchange format, while being suitably comprehensive in documenting the data relevant to the campaign. In the
context of international data sharing of substratum and benthic spectral data, the establishment of standards for the capture,
storage, and use of spectral signature files with associated metadata is required due to the effect of environmental factors in
shallow water environments on the derived data (Dekker et al., 2010). The standard proposed in Section 2 can be easily
implemented as a schema in a common exchange format such as GML and XML (Extensible Markup Language). XML is
selfdescriptive with extensibility features (Mahboubi and Darmont, 2010) and can facilitate progress towards integration of in situ
coral reflectance data with multi-dimensional remote sensing data sets, both within the marine context and near-shore terrestrial
campaigns. One of its greatest strengths is platform independence, and a framework for XML-based data interchange is espoused
in the Common Warehouse Metamodel, which includes XML Metadata Interchange (XMI) standards for datawarehouses
(Mangisengi et al., 2001
Large-scale implementation of standards for encoding and sharing coral reflectance metadata is best facilitated by national and international agencies responsible for safeguarding and distributing these datasets. OGC (Open Geospatial Consortium) launched the Marine Metadata Interoperability Project to make data available from various ocean observing systems (OGC, 2012); however there are no specific metadata standards for in situ marine spectroscopy. IMOS (Integrated Marine Observing System, Australia) provides NetCDF specifications for in situ marine observations but are biased towards biochemical sensors and recording environmental variables, with no reference to spectroscopy measurements (IMOS, 2012). The ISO19115 sets of standards for geospatial metadata provide general guidelines, but do not explicitly address the metadata requirements of marine field spectroscopy collection techniques, or the ontologies and data dependences required to model the complex interrelationships among the observed phenomena as data and metadata entities (ISO, 2012). The lack of international standards impedes wide-scale mining and sharing of in situ marine spectroscopy datasets generated by remote scientists around the world. Adopting an XMLbased metadata model for coral reflectance measurements is an initial step in establishing the foundations for a standard.
A practical and viable metadata standard for in situ coral reflectance can be used to inform a common data exchange standard for spectroscopy datasets in general. The model presented in this paper meets the requirements for a metadataset that is comprehensive, explicit, allows replication of the campaign if required, and is suitably broad in the scope of data capture to permit interoperability with other datasets. The standard is flexible by specifying both critical and useful metadata fields that are populated dependent upon the logistics of the campaign and the purposes for which the data will be used. In situ spectroscopy metadatasets are currently generated based on ad hoc data collection protocols that impede wide-scale data mining, sharing, intercomparison and interoperability of datasets. A metadata model based on the standard proposed here, in a common exchange format such as XML would facilitate convenient and practical data exchange among the remote sensing community.
Marine remote sensing scientists at the ACEAS (Australian Centre for Ecological Analysis and Synthesis) Bio-optical workshop held in Australia in 2012 who generously provided input to the coral reflectance metadata schema proposed here. A. G. Dekker, V. E. Brando, J. M. Anstee, A. J. Botha, Y. J. Park, P. Daniel., T.J.M. Malthus, S. R. Phinn., C. M. Roelfsema, I.A. Leiper, S. Fyfe, “A Comparison of Spectral Measurement Methods for Substratum And Benthic Features in Seagrass and Coral Reef Environments” Proceedings of ASD and IEEE GRS; Art, Science and Applications of Reflectance Spectroscopy Symposium, Vol. II, 15pp, Boulder, CO, 2010.
M.J. Duggin, "Factors limiting the discrimination and quantification of terrestrial features using remotely sensed radiance", , International Journal of Remote Sensing, 6: 1, 3-27, 1985.
IMOS, “IMOS NETCDF FILE NAMING CONVENTION”, Version 1.4, February 22 2012, viewed October 14, 2012 http://imos.org.au/fileadmin/user_upload/shared/IMOS%20General/documents/Facility_manuals/IMOS_netCDF_filenaming_con vention_v1.4.pdf ISO, "ISO 19115:2003 Geographic information -- Metadata", 2003, viewed October 14 2012, http://www.iso.org/iso/catalogue_detail.htm?csnumber=26020 H. Mahboubi and J. Darmont, "Optimization in XML Data Warehouses", pp 232-253 E-Strategies for Resource Management Systems: Planning and Implementation, E. Alkhalifa (Ed.), University of Bahrain, 2010.
T. Malthus and A. Shirinola, "An XML-based format of exchange of spectroradiometry data", EARSeL Imaging Spectroscopy
SIG, Tel
O. Mangisengi, J. Huber, C. Hawel, W. Essmayr, "A Framework for Supporting Interoperability of Data warehouse Islands Using XML", Lecture Notes in Computer Science, 2001 Data Warehousing and Knowledge Discovery, Volume 2114, 328-338, 2001. OGC, "Marine Metadata Interoperability Project", viewed October 14 2012, ttp://www.ogcnetwork.net/node/345 K. Pfitzner, R Bartolo, G Carr, A Esparon & A Bollhöfer , “Standards for reflectance spectral measurement of temporal vegetation plots”, Supervising Scientist Report 195, Uniprint NT, Darwin, 2011.
C. Roelfsema, J. Marshall, E. Hochberg, S. Phinn, A. Goldizen, and K. Joyce, “Underwater Spectrometer System 2006 (UWSS04)”, University of Southern Queensland, 2006, viewed August 01 2011, http://ww2.gpem.uq.edu.au/CRSSIS/publications/UW%20Spec%20Manual%2029August06.pdf.
R. Torlone, “Encyclopedia of Database Systems”, Part 9, p. 1560-1564, Springer Science + Business Medi
Barbara Rasaiah is a PhD candidate at RMIT University in Melbourne, Australia, investigating approaches to a coordinated evolution of hyperspectral metadata protocols, field spectroscopy methods and data exchange standards within the hyperspectral remote sensing community. Barbara’s work has been presented at the ISRSE 34 conference, 7th EARSeL workshop, and ISPRS 2012. Barbara has an educational background in computer science and mathematics and has worked in industry as a computer programmer, web designer, and computer operations analyst. She was awarded the 2012 Goetz Instrument Award from ASD Inc., for novel and innovative research in field spectroscopy.
Simon Jones is professor of remote sensing and director of the Remote Sensing and Photogrammetry Research Centre at RMIT University in Melbourne, Australia. His current projects include leading research at TERN (Terrestrial Ecosystem Research Network), Commonwealth Environment Research Fund Hub “Landscape Logic”, and organising the 2012 ISPRS International Congress on Photogrammetry and Remote Sensing. Simon’s specializes in remote sensing, ground verification (in situ observations), spatial analysis, spatial data uncertainty, land-cover mapping, monitoring & modelling and vegetation. He is a foundation member and former director of the (Surveying and) Spatial Sciences Institute, Australia and has previously worked at the Joint Research Centre of the European Commission (Global vegetation Monitoring Unit).
Chris Bellman is associate professor and discipline head of geospatial science at RMIT University in Melbourne, Australia. His current projects include organising the 2012 ISPRS International Congress on Photogrammetry and Remote Sensing. Chris specializes in photogrammetry, GIS and spatial analysis and computer-aided mapping. He is a previous president of the Surveying and Spatial Sciences Institute of Australia. Chris is winner of the 2008 Victorian Spatial Excellence Award for Education and Professional Development.
Tim Malthus is leader of the Environmental Earth Observation program in CSIRO Land and Water in Canberra, Australia. His current projects include TERN (Terrestrial Ecosystem Research Network), IMOS (Integrated Marine Observing System) and the investigation of land use and land cover classification at high resolution. Tim’s specialization in calibration/validation activities, and field spectroscopy with analysis of airborne and satellite Earth observation data, is applied in the development of improved monitoring tools for informing wider environmental policies. He has held positions as Senior Lecturer in Remote Sensing, University of Edinburgh, 1994–2009 and Director of the NERC Field Spectroscopy Facility, UK, 2004–09.