R.D. Graetz, A.J. Prata, J.S. Wallace, and I.J. Barton
The compilation of this document was both a pleasure and a challenge.
It was a pleasure because all the authors believe in the concept of the Earth Observation Centre and are committed to its implementation. It was challenge to find and articulate the shared core of agreement. That was not easy: the debate has been spirited and detailed with independent inputs from many interested research staff.
Even so, the four authors of this document share its vision and mission, and are committed to their implementation. However, the separate groups of research staff represented by each author differed in their support of the three areas of research activity defined below. Research areas 1 & 2, which will account for almost all activity (>95%) was supported everywhere, whereas area 3 was only supported by Melbourne, Hobart and Canberra.
In this section of our report, we set out the argument that we developed since being commissioned and which has shaped the research strategy proposed in the next section.
Two propositions underpin all.
The first is that CSIRO is a research organisation: its prime directive is to undertake scientific research. Within CSIRO, components such as COSSA and CIRC, have different functions. Even though the concept of the Earth Observation Centre arose from a COSSA-initiated review, and COSSA has the responsibility to create the Earth Observation Centre, the essential nature of the Earth Observation Centre will differ substantially from that of COSSA. The Earth Observation Centre will be part of mainstream CSIRO. Therefore the prime directive and ethos of the Earth Observation Centre must be the conduct of scientific research.
This is an important point that was raised in the Implementation Document (COSSA Steering Committee) and we reiterate it here.
The concerns that led to the concept of the Earth Observation Centre partly arose from data issues; of these, some were related to research but most were not. A focus on data issues dominates the circulated vision and mission discussions in the 'Implementation Plan' and the draft 'Outline of Strategic Plan, etc.'. We do not accept this emphasis. We are convinced that an Earth Observation Centre within CSIRO must be based on relevant, high quality research activities.
Therefore, this document proposes a wider view that embraces all data issues and integrates them within a comprehensive research strategy.
This view is that in the current era of Earth observation activities, research and data are inextricably and synergistically linked.
It is obvious to us that a strategy of emphasising data at the expense of research has no future in the CSIRO of today or tomorrow. Conversely, a strategy of research that ignores data management will continue the emasculated conditions of these last few years that led to the formation of MDP # 19, and its subsequent review.
The second proposition is that the Earth Observation Centre should address the big issues: those characterised by broad spatial scales over long time intervals to which Earth Observation data make a unique and underpinning contribution. These are the issues that can really only be addressed with space data. Moreover, we should address not only the current research questions but what is more important, consider those questions that are not presently being asked.
The Earth Observation Centre should become a leading and singular research institution.
Furthermore, we believe this can be achieved in a manner that is complementary to activities within Divisions and without excluding any significant research themes. We illustrate this with a hypothetical research program in the Appendix.
On the basis of on these two propositions, the central thrust of this document is to propose a bold but operationally sustainable integration of research and data management, the Research Strategy, to underpin the Earth Observation Centre over the next five years, 1995-2000.
In addition, we also discuss several implementation issues that we believe are not usefully separated from the proposed Research Strategy. Like the Strategy, these implementation issues have been discussed amongst us and the views of our colleagues that were canvassed and considered.
To illustrate and support our proposals made in the next section, we here collect some the many facets of shaping the Earth Observation Centre that were considered.
The acceptability of the Earth Observation Centre within CSIRO management and its attractiveness to research staff will be determined almost entirely by the formulation and implementation of its Research Strategy.
Therefore, once accepted, the Earth Observation Centre Research Strategy must be widely and speedily communicated within CSIRO, and implemented in the Earth Observation Centre development.
Because COSSA and the Earth Observation Centre have very different roles and responsibilities within CSIRO, the Earth Observation Centre must be independent of COSSA in the pursuit of its role within the constraint of its essential requirement to report to the Director of COSSA.
Therefore, it is essential that the research and development directions of the Earth Observation Centre be set by the Leader and staff with input from Divisions and an Advisory Committee (see later).
Now that the Earth Observation Centre is a reality, albeit still a fledgling, its identity and support needs to be indelibly confirmed.
Therefore, we recommend that MDP #19 be terminated as soon as is possible and that its remaining funding be transferred to the Leader of the Earth Observation Centre for disposition, and that this financial independence be continued;
in addition, we believe it to be critical that the Earth Observation Centre creates and communicates its own identity, and manages its own information access independently of COSSA.
Two aspects of the creation of the Earth Observation Centre will enhance remote sensing within CSIRO: by providing a disciplinary focus, and, more influential, by the collocation of staff. However, if staff remain distributed across Australia, then opportunities for productive interaction will remain limited and thus the output of the Earth Observation Centre is unlikely to be very different from the current output from the separate Divisions.
Therefore, the collocation of staff is obviously the preferred option wherever possible but it remains a practical problem that must be faced.
The Earth Observation Centre research areas should not be traditional discipline areas, for otherwise the overlap with Divisions will be too great and the necessity of the Earth Observation Centre will be questioned.
Therefore the Earth Observation Centre must include in its research focus, areas that do not rest comfortably within Divisions or cover areas common to many Divisions.
While it is important to set out the long term strategic goals of the research and development, their creation will always risk the generation of unrealistic expectations amongst some users.
Therefore, the Earth Observation Centre should set both long term strategic goals as well as short term objectives arising from interactions with the user communities;
the Earth Observation Centre objectives should be clear, open and achievable with the resources available.
The research strategy that we outline below is simple, general, and has three components; vision, mission and activities.
No specific objectives are defined.
We see the essential role of the Earth Observation Centre as being synergistic and facilitating for research scientists already actively engaged in their own research and responsive their own interests, capabilities and funding. The focus of the Earth Observation Centre will enhance this research activity, and through coordinated data management, benefit a larger and more diverse clientele than before.
The focus of the Earth Observation Centre is not to re-orient existing research in the short term: rather, we want to do what we do far more effectively.
We begin with a succinct statement of the essential nature and direction of the Earth Observation Centre:
'To ensure that Australia gains maximum benefit from the application of Earth Observation data.'
Translating the Vision statement into a broad action plan, we write:
'The Earth Observation Centre will conduct multidisciplinary research and development based on remotely sensed and other data for the benefit of the Australian and global communities'
We see this benefit being realised from the innovative and efficient extraction of relevant information from Earth Observation data. These three adjectives link research, development and application; they describe the three components required for an end-to-end approach.
The research objectives pursued within the Earth Observation Centre will be diverse, reflecting the interests and collaboration of individual scientists. Therefore, as discussed above, it is difficult and counter-productive to define and prioritise research objectives. It is far more useful to define the three areas of research activity within which the spectrum of research effort could be managed. Continuing and potential cooperation and collaboration with research staff from CSIRO Divisions and other organisations (DEST, DoPI&E, DSTO) is implicit throughout.
While these activities are broadly defined below, they proscribe farm-level decision-support systems, that are increasingly being covered by other public and private sector organisations.
The three areas of research activities follow. These activities are exemplified in the Appendix that also illustrates the nature and extent of the potential interactions between the EOC and CSIRO Divisions.
1. Earth Observation Science and Data Analysis: To undertake the research necessary for the development of generic methods (measurement models) to relate satellite data recorded above the Earth's atmosphere to geophysical variables; these variables being required to address broader current and potential research questions within the Earth Observation Centre, CSIRO and the wider research community.
Measurement models include scene, sensor, atmosphere and image processing sub-models that in combination, determine the accuracy and precision of the derived geophysical variables.
Measurement models are required for the retrospective analysis of current Earth Observation data archives, eg. AVHRR, as well as for the prospective datasets, eg. ADEOS, SPOT Vegetation Instrument, SeaWIFS and EOS.
While it has and will continue to be applications research that creates the need for highest quality Earth Observation datasets, it is the development and application of measurement models that remains the key to the production of the value-added datasets required by the research community.
As an illustration, an AVHRR Science Working Group has been established, originally within MDP #19 and now the EOC, to prescribe the processing of a standard set of geophysical datasets to be generated by the EOC and made available to CSIRO research staff: these datasets include sea surface temperature (SST), land surface temperature (LST), surface albedo (a VIS, a NIR, a SOLAR) and NDVI (a vegetation index). In addition, an analogous GMS Pathfinder Science Working Group has been established in collaboration with the University of California, the Bureau of Meteorology, the Australian National University, and the University of Tasmania.
2. Calibration / Validation: To undertake the required field and laboratory research to support the development of measurement models evaluating either the accuracy of sensor performance (calibration) or of derived geophysical variables (validation).
Both measurement models, when translated to algorithms, and Calibration / Validation activities as a source of ancillary data and process understanding, are fundamental to the generation and documentation of geophysical data sets for distribution by the CSIRO DAACs for subsequent application by the wider research community.
Calibration / Validation research activities are critical to success in addressing many contemporary scientific questions: it is not an unreasonable assertion that it is now the most critical issue in the application of space data. The importance of Calibration / Validation research is now widely recognised by organisations such as CEOS and EOS, and by individual spacecraft missions that have sought, or are seeking CSIRO expertise in this research area; eg. the ATSR instrument on ERS-1/2, POLDER on ADEOS, and the Vegetation Instrument on SPOT 4.
Calibration / Validation research activities will ensure the highest quality datasets are available to the Australian and global research communities. It is anticipated that within this research activity, there will be considerable collaboration with Divisional research staff; see Appendix.
3. Sensors and Systems: Australia has a growing aerospace industry that has, in the past, found valuable research support for development programs from within CSIRO. These activities will continue to grow and CSIRO has a vital role to play in supporting these programs. The research skills required for the success of development programs in Earth Observation are currently not available from other Australian institutes.
It is important that there is a national focus for this area of research and, for CSIRO, the activity is far better suited to the EOC than to individual Divisions.
The Earth Observation Centre can become this focus through secondment from CSIRO or from industry, and by managing the research component only.
These three research activities represent a significant change in emphasis and allocation of tasks from the Earth Observation Centre concept proposed by COSSA in the draft Implementation Plan. The emphasis has been placed on getting the research component right but not to the exclusion of data management that is implicit in all that is stated above.
Co-ordinated scientific research and development is the only sustainable locomotive for generating useful Earth Observation value-added products;
Strategic research and development first requires, and then generates Earth Observation datasets;
The two key elements are co-ordinated Research and Development, and Data Management.
Both elements are critical and interactive, but for the Australian community to benefit from Earth Observation Centre activities the nature and direction of Data Management activities must support and complement those of Research and Development.
We envisage an interactive interdependence of Research and Development and Data Management within the Earth Observation Centre, and this relationship is simplified in the following diagram.
Data Management alone is a relatively simple problem to solve. It is far more difficult to co-ordinate Research and Development with Data Management so that end-to-end solutions are provided and the value-added datasets generated and captured. Nonetheless, we see the solution to this problem as being one of research-leadership rather that data-driven.
Because the scope of Earth Observation Centre research activities is so broad, it is unlikely that any one person can be an effective research leader. The maintenance of research standards can be assured with the review process within and without the Earth Observation Centre. However, the maintenance and co-ordination of research effort and the interaction with clients, both actual and potential, are by far the most important tasks to keep the Earth Observation Centre on track and viable.
Thus, we see the critical attributes of the Leader as being:
We also believe the next critical step in the development of the Earth Observation Centre is the appointment of the Leader. We urge that this be urgently addressed with the leader being appointed for an initial period of 5 years, located in Canberra.
REFLECTION and GUIDANCE
The Earth Observation Centre is a new concept within CSIRO and effectively without peer within Australia; the organisations of ERIN (DEST), NRIC and AGSO, both (DoPI&E), being considered in this assessment. The role models, or at least analogues of the Earth Observation Centre are thus mostly international. Therefore, attention must be given to providing mechanisms that will provide an international reflection of our direction, activities and performance in a way that will support the best possible management of the Earth Observation Centre resources; people, funds, and time.
Two mechanisms are suggested: the Earth Observation Centre Fellowships, and an Advisory Committee.
Earth Observation Centre Fellowships
The Earth Observation Centre Fellowships have already started with the first and current award being to Jim Simpson. Jim Simpson has played a significant role in the concept and establishment of the Earth Observation Centre and he has continued to contribute to research during the brief period of his Fellowship.
Following in this role, Earth Observation Centre Fellowships can provide a mirror to our performance and direction. To do this, the award needs to be to a distinguished but interesting and active scientist, for a brief period of time. Earth Observation Centre Fellowships are different from, and in addition to, visiting research scientists to the Centre.
It is recommended that the award by invitation of Earth Observation Centre Fellowships be continued at an annual rate, and that they continue to be carefully selected such that the benefit to the Earth Observation Centre is clear and widely distributed.
We believe that the direction and functioning of the Earth Observation Centre would benefit from interaction with a small expert committee - an Advisory Committee.
Because of the vision and mission of the Earth Observation Centre, we suggest this small committee be convened annually and comprise the following:
The intent and function of this Advisory Committee are clear from our suggested membership. The final choice should be made by the Leader of the Earth Observation Centre.
We believe it is critical to regularly take professional advice from a wide range of sources; particularly from outside CSIRO and Australia because the Earth Observation Centre will become a significant functional link in the emerging international web of Earth Observation science and data analysis. To orient our activities by the Earth Observation Centre Vision - to ensure that Australia gains maximum benefit from the innovative and efficient application of Earth Observation data - we must regularly assess our performance by international standards.
Research Strategy exemplified
To illustrate how the research strategy would be applied in the EOC, a fictitious example is described which also illustrates the perceived roles of the EOC and individual Divisions and the way these groups may interact on common research topics.
The example given is for an application of AVHRR data, but future EOC activities will also include data from other earth observation instruments and platforms (e.g. SAR, SeaWiFS, OCTS, MODIS etc.).
THE APPLICATION OF AVHRR DATA IN THE MANAGEMENT AND RESEARCH OF LOCUST PLAGUE EVENTS
1. Reception (direct or indirect) of raw AVHRR data over the study area.
2. Routine processing of AVHRR data to provide at-satellite, navigated brightness temperatures in the infrared channels and radiances in the visible/near infrared channels.
This activity requires conversion of the raw data using calibration coefficients and navigation into earth coordinates using orbital elements, an orbital model and ground control points.
3. Derivation of standard geophysical products from the brightness temperatures/radiances.
These products include sea surface temperature, land surface temperature, surface albedo and NDVI (a vegetation index).
The production of these parameters requires the detection of cloud contaminated data, correction for atmospheric effects (absorption and scattering) and the inclusion of surface and solar geometry.
The four products mentioned above should be considered standard products that will be produced automatically by the EOC for use by CSIRO Divisions in their research programs, and in applications developed by particular divisions.
4. Standard statistical techniques for data analysis, compositing of data, production of time series, filling in missing (cloudy) values, etc.
5. Development of new parameters for highlighting locust active or affected areas (new NDVIs?, thermal data may be useful here?).
Comparisons with climatological data.
Collection of ground data for validation of satellite-derived parameters.
Development of task-specific analysis techniques.
6. Assessment of locust damage, forecast of locust paths, deployment of ground based resources (aircraft spraying of insecticide etc.).
Use of satellite data to forecast future areas of locust plague outbreaks.
EOC activities: Actions 1, 2, 3 and 4 as described above are activities that are common to any research project or application of satellite data and these should be seen as the responsibility of the EOC. The first research activity described in the strategy thus includes investigations to provide high quality standard geophysical data products from satellite data. These activities will include data calibration, cloud detection and clearing, correction for atmospheric effects in the visible, infrared and microwave bands, navigation of satellite data and the development of common statistical techniques for data analysis.
Divisional activities: Actions 5 and 6 clearly belong in individual Divisions of CSIRO as they pertain to specific research programs and applications. However, there are obviously areas here where some interaction is required. For example, the ground validation data collected by the Division should be used by the EOC in the assessment of the quality of their derived standard products. Divisions may also call on the particular skills of the EOC for the joint development of new products or for specific analysis techniques that may be required for their particular application.
Copyright CSIRO 1999 ©
Use of this web site and information available from it is subject to our
Legal Notice and Disclaimer