4

Data Systems

A sustained land imaging program pays close attention not only to data acquisition, but equally to data management, data products, and data availability. Satellite data, which were once largely inaccessible and required specialized technical infrastructure to manipulate images, have evolved over the past few decades. Today, many satellite products are freely and openly available, usually via the Internet, and easily accessed by commercial and open-source software. An array of products far beyond simple imagery has been produced, such as topography, land cover, vegetation class, and vegetation performance (productivity, water use, phenology, and other attributes). For more technical users, information about algorithms, uncertainty, and ground truth are usually available. In envisioning future land imaging, end to end, the documentation of data transformations (data product generation), tracking of uncertainty, execution and documentation of calibration and validation activities (including ground truth), and planning of data availability are also core activities.

DATA AVAILABILITY

The policies for data availability have changed dramatically over the life of the Landsat program from a fee-for-service model to the U.S. Geological Survey’s (USGS’s) current open-access policy via the Internet. In the best sales year, approximately 25,000 images were sold. The Landsat data distribution now exceeds that number in a single day (Figure 2.1).1

Freely available data from USGS—not only Landsat data, but also airborne imagery and data on topography, hydrology, land cover, and so on—are widely downloaded and applied to scientific research and resource management. A number of products based on Landsat are available, along with a rich set of related map and imagery-derived products. The USGS Earth Resources Observation and Science (EROS) Center identifies several hundred current digital map and imagery products. These products span a variety of themes and mostly comprise aerial and satellite imagery, with a few dozen map products derived from imagery.

The EROS Center operates at least seven sites for downloading moderate- and high-resolution imagery and related geospatial data. GloVis2 allows users to retrieve data in two or three steps through an interactive interface; EarthExplorer3 provides access to many more data sets, although the search engine is less intuitive; the National

____________

1 See http://landsat.usgs.gov/mission_headlines2012.php.

2 See http://glovis.usgs.gov/.

3 See http://earthexplorer.usgs.gov/.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 39
4 Data Systems A sustained land imaging program pays close attention not only to data acquisition, but equally to data man- agement, data products, and data availability. Satellite data, which were once largely inaccessible and required specialized technical infrastructure to manipulate images, have evolved over the past few decades. Today, many satellite products are freely and openly available, usually via the Internet, and easily accessed by commercial and open-source software. An array of products far beyond simple imagery has been produced, such as topography, land cover, vegetation class, and vegetation performance (productivity, water use, phenology, and other attributes). For more technical users, information about algorithms, uncertainty, and ground truth are usually available. In envisioning future land imaging, end to end, the documentation of data transformations (data product generation), tracking of uncertainty, execution and documentation of calibration and validation activities (including ground truth), and planning of data availability are also core activities. Data Availability The policies for data availability have changed dramatically over the life of the Landsat program from a fee- for-service model to the U.S. Geological Survey’s (USGS’s) current open-access policy via the Internet. In the best sales year, approximately 25,000 images were sold. The Landsat data distribution now exceeds that number in a single day (Figure 2.1).1 Freely available data from USGS—not only Landsat data, but also airborne imagery and data on topography, hydrology, land cover, and so on—are widely downloaded and applied to scientific research and resource manage- ment. A number of products based on Landsat are available, along with a rich set of related map and imagery- derived products. The USGS Earth Resources Observation and Science (EROS) Center identifies several hundred current digital map and imagery products. These products span a variety of themes and mostly comprise aerial and satellite imagery, with a few dozen map products derived from imagery. The EROS Center operates at least seven sites for downloading moderate- and high-resolution imagery and related geospatial data. GloVis2 allows users to retrieve data in two or three steps through an interactive interface; EarthExplorer3 provides access to many more data sets, although the search engine is less intuitive; the National 1  See http://landsat.usgs.gov/mission_headlines2012.php. 2  See http://glovis.usgs.gov/. 3  See http://earthexplorer.usgs.gov/. 39

OCR for page 39
40 LANDSAT AND BEYOND Map Viewer and Download Platform4 contains data available via EarthExplorer but also contains data from the National Agriculture Imagery Program. LandsatLook5 is a map-based interface where a user can search scene avail- ability and view candidate scenes. Web-Enabled Landsat Data (WELD6) are obtainable from yet another website and include atmospherically corrected Landsat images. Landsat data are available from Landsat.org, 7 which is operated by Michigan State University, and from the Global Land Cover Facility at the University of Maryland. 8 Moreover, several commercial companies also serve high-resolution aerial and spaceborne images, ­ andsat L imagery, and products based on imagery (e.g., ESRI, Google, Microsoft, and Yahoo). While these sites and ser- vices offer innovative ways to search for, display, and provide images and products based on them, they lack the comprehensive access to land imaging archives that can only be offered to the public from an authoritative federal government source. These programs and others like them could be better integrated to form the basis for a ­ oherent c land imaging program. Benefits of the current open-access policy are significant and have allowed use of the federal investment in Landsat by a vastly larger user base, including all sectors—from basic research, land management research and applications, education, citizen use, and use by the value-added sector. Maintaining open access is critical. Moving toward the future, the use of land imagery can be further increased, and additional value can be gained by enhancing the suite of data products, improving their documentation through metadata and uncertainty tracking, and developing even more advanced data discovery and distribution channels. Products Derived from LAND REMOTE SENSING The Landsat series of satellites provides the required long-term continuity of imaging for scientific and soci- etal benefit purposes. However, the Landsat sensor, by its nature, cannot provide all information required for land science and management. Investment in new data products must be balanced between additional advanced data products from Landsat and new data products from other emerging data sources, such as airborne LiDAR and other airborne and spaceborne sensors. Large quantities of novel data are being collected: critical near-term decisions will need to be made about investment levels to access, process, document, and distribute them. The Sustained and Enhanced Land Imaging Program (SELIP) will benefit from an effective user-oriented mechanism, through advisory committees or other structures, to prioritize different data sets and evaluate the relative importance of enhanced data products from legacy sensors compared to new techniques. There is potential for a far greater array of derived products than are currently available. If appropriately defined and funded, sustained land imaging capabilities would enable a myriad of products and services, including many essential climate variables and climate data records. Most of the products would be difficult for users to code themselves. The complexity of the transformations needed to render some observations into useful products—which in extreme cases are millions of lines of code requiring high-performance computing—makes better infrastructure imperative. With the availability of baseline products, the population of users would also expand, driving demand for successively higher level products. The situation is not unlike the supply of “app” products for cell phones; however, without a sustained land imaging program, the product stream will diminish. As part of an evolving imaging system, SELIP could identify critical data products and drive requirements for future missions. Because the knowledge and technology needed to produce land-surface information from imagery are sometimes formidable, it makes sense to provide such information from a data system rather than require users to undertake the transformations. Focusing on specific data products can add a great deal of rigor to the requirements definition process for follow-on missions. Management and funding models are part of ensuring that the products are produced, validated, and available for use. USGS already distributes valuable data products derived from land imagery—for example, the National Land Cover Dataset, LANDFIRE, the Global Land Survey, and Land Surface Reflectance. The Land Surface Reflectance 4  See http://nationalmap.gov/viewer.html. 5  See http://landsatlook.usgs.gov/. 6  See http://weld.cr.usgs.gov/. 7  See http://landsat.org/. 8  See http://www.landcover.org.

OCR for page 39
DATA SYSTEMS 41 product is available for the Global Land Survey 2000, 2005, and 2010 collections and is generated on demand from Landsat 4-5 TM and Landsat 7 ETM+ data.9 The concept would gain additional utility through a formal but open mechanism for identifying candidate products and the resources needed to produce them. Although the concept of a climate data record (CDR) has surfaced numerous times in recent National Research Council reports,10 the climate research and policy communities continue to struggle with an exact approach to meet this need (i.e., one that is both sufficient and cost effective). In addition, satellite-based CDRs have been further segmented into the following: • Fundamental climate data records (FCDRs) are calibrated and quality-controlled sensor data together with documentation for the data used to calibrate them. • Thematic climate data records (TCDRs) are geophysical variables derived from the FCDRs that have well- defined levels of uncertainty, with an ongoing program of correlative in situ measurements required for validation. • Essential climate variables (ECVs) are atmosphere, ocean, and land measurements derived from FCDRs and TCDRs. They have to be technically and economically feasible for systematic observation and sufficient to meet the needs of the United Nations Framework Convention on Climate Change and the Intergovernmental Panel on Climate Change (IPCC). To be useful, the ECVs must be a time series with sufficient length, consistency, and continuity to identify climate variability and change. This report has looked at observations that would be available from SELIP from the perspective of the needs of users and engineering units (spectral, radiometric, spatial and temporal resolutions, and so on). The com­ ittee m recognizes both the challenge and the need for SELIP to work with key related communities to develop an agreed- on set of FCDRs, TCDRs, and ECVs based on moderate-resolution sensors. That will mean going beyond the engineering units, such as calibrated radiance, in the existing Landsat archive, embracing more broad units—such as surface reflectance, surface temperature, cloud, and cloud shadow—and eventually evolving to more application- oriented products (i.e., ECVs). These products will also need to meet the Global Climate Observing System, First National Climate Change Communication, and IPCC requirements and be technically and economically feasible systematic observations. The Landsat Surface Reflectance product is an excellent example; it is produced routinely for selected time periods but is also available on demand for specific Landsat scenes. Algorithm Development and Calibration/Validation As the focus in Landsat and other space or airborne data acquisition systems evolves from providing imagery to providing higher-level data products derived from those images, a set of consequent activities becomes neces- sary. The first step is to develop a rigorous process for determining the required data products, similar to NASA’s elicitation of requirements for Moderate Resolution Imaging Spectroradiometer (MODIS) data products or the development of the essential climate variables. Once a set of desired standard data products has been determined, the algorithm for producing the data product is selected, reviewed, and implemented. Models for this process exist in the federal and private sectors—ranging from proprietary development in house to open, competitively selected development. Regardless of the model, the selection, development, and distribution of algorithms are best achieved with freely and openly available data. Transparency of algorithms provides credibility and allows a larger com- munity to participate in evaluation and continuous improvement. 9  U.S. Geological Survey, Product Guide: Landsat Climate Data Record (CDR) Surface Reflectance, Version 2.0, 2013, available at http:// landsat.usgs.gov/documents/cdr_sr_product_guide.pdf. 10  See the following National Research Council (NRC) reports: Climate Data Records from Environmental Satellites (2004); Adequacy of Climate Observing Systems (1999); Ensuring the Climate Record from the NPOESS and GOES-R Spacecraft: Elements of a Strategy to Recover Measurement Capabilities Lost in Program Restructuring (2008); Earth Science and Applications from Space: National ­ mperatives I for the Next Decade and Beyond (2007). Each report was published by ­ ational Academy Press (after mid-2002 The National Academies N Press), Washington, D.C.

OCR for page 39
42 LANDSAT AND BEYOND An instrument flown in a sustained land imaging program has a requirement to produce calibrated radiances. Calibration and validation of data products is critical for their effective use and credibility. 11 A strength of the Landsat program has been the radiometric calibration of the instrument, along with spatial and temporal compari- sons. The development of rigorous data products requires both onboard instrument calibration and comparison to well-known ground targets. Images without rigorous calibration support limited analyses, but the associated data will not support higher-level products. An ongoing process of instrument evaluation provides validation of radio- metric data products, such as reflectances, and is a basis for validation of high-level data products. However, as quantitatively derived products, such as topography, land cover, or leaf area, are developed, these products too are based on a careful and systematic program of calibration and validation against measurements made on the ground, by aircraft underflight, and by other means. The results from these calibration/validation programs contribute to credibility and are most useful when they are openly available with the data. Findings Freely available data from the Landsat program have brought enormous benefits to science and to operational users.12 Higher-level products continue to be developed, providing ever greater benefits to society at large. USGS websites and other venues effectively provide access to imagery and derived products, with varying degrees of ease of use. However, the hierarchical organization and plethora of websites and interfaces make access difficult, especially for novice users who may not know which data are on which sites. The government currently uses a number of approaches to distribute Earth observation data: dedicated federal data centers, data federations such as the Earth Science Information Partners, commercial value-added resellers, and Internet information distributors in the private and nonprofit sectors. All these mechanisms could be used in assembling an infrastructure for the SELIP, as long as primary data and key data products remain available under an open data policy. The potential list of baseline products and services that land imaging could provide is much larger than the suite of products and services currently provided. However, (1) the mechanisms and procedures for introducing change are cumbersome in all agencies, so the user community cannot realistically implement new products or new algorithms for existing products; (2) similar products from NASA and NOAA are global in scale and are produced whenever and wherever the input data are available, regardless of demand; and (3) the private sector supplies some derived products of varying quality and degrees of validation. Recommendations USGS, as part of the Sustained and Enhanced Land Imaging Program, should continue to deliver derived products from imagery without explicit cost to the end users. USGS should • Improve search capabilities and transparency to users and • Continue to interface with the private sector to improve access to public- and private-domain land imaging data products and services. The Sustained and Enhanced Land Imaging Program should develop a systematic process for identify­ ing and prioritizing a wider suite of products, including essential climate variables, that can be derived 11  The international Committee on Earth Observing Satellites has advocated a universal validation data set for all global land cover products to increase the interoperability of data from many countries’ satellites. It also emphasizes validation and accuracy assessments as a major part of a mapping program. Strahler et al., Global Land Cover Validation: Recommendations for Evaluation and Accuracy Assessment of Global Land Cover Maps, 2006, available at http://nofc.cfs.nrcan.gc.ca/gofc-gold/Report%20Series/GOLD_25.pdf. 12  National Geospatial Advisory Committee-Landsat Advisory Group Statement on Landsat Data Use and Charges, September 18, 2012, available at http://www.fgdc.gov/ngac/meetings/september-2012/ngac-landsat-cost-recovery-paper-FINAL.pdf.

OCR for page 39
DATA SYSTEMS 43 from moderate-resolution land imagery, and for documenting and validating algorithms, including their modifications or replacements. In doing so, the program should • Define criteria that government-provided authoritative data sets should meet, among them such attributes as calibration, accuracy assessment, and validation, and including ground truth; • Define criteria for which products should be provided by the government and which by the private sector; • Implement procedures for development, cost estimation, peer review, and the publication of algo- rithms that produce derived products; and • Implement plans, procedures, and budgets for ongoing validation.