David L B Jupp
The formation of the Earth Observation Centre (EOC) is a very welcome development. It is a concept that, in various forms, has been proposed and suggested often in CSIRO over the last 15 years but has always eventually fallen on barren ground. The Simpson Report provided a catalyst for an EOC to finally come into being. However, the EOC would not have progressed to its present point if there were not an underlying and widely accepted desire to see something of this form develop. I believe that the desire existed before the Report was commissioned and involves much more than the data collection, storage and communication focus of MDP-19. Not withstanding this desire, the EOC can still be difficult to maintain. This will occur unless there is a continuing level of strong commitment and support from the Organisation and from the individuals on whose skill and proud accomplishments its future success rests. Equally, with such support and commitment it cannot fail.
Over those same 15 years there have developed strong centres of remote sensing expertise in CSIRO. These have mostly had an applications focus and been located in a range of different industry oriented Divisions. Decentralisation has certainly lead to some of the difficulties listed in the Simpson Report. However, on the positive side, the period has seen remote sensing science in CSIRO develop soundly and productively in support of well defined application areas. These include atmospheric sciences, fisheries and oceans, water resources, mineral and geophysical exploration, soil survey, forestry and agricultural applications. The integration of the remote sensing scientists with applied Divisions has been very fruitful in a way that their separation into a single disciplinary Unit in the past may not have been. Australia has become very well known for its application oriented combination of remote sensing science and real world applications.
There has also developed strong support for a science focussed EOC from these groups. This has occurred in no small way because, in the words of the Simpson Report, "a given group's work attempts to cover instrumentation development, ground station operations, algorithm development, applications development, validation and product delivery. This task is too big for small, geographically isolated groups." The Simpson Report arose from a review of the Data Acquisition and Utilisation MDP-19. However, I believe the strong move towards an EOC in CSIRO does not come from needs unique to the data gathering segments but from the wider truth of this statement. It applies to the whole of the productive remote sensing and Earth Observing research community that has developed in CSIRO.
The time has come when the value of focussing the skills of groups of scientists on issues common to the remote sensing science underpinning the applications across many Divisions has become too great to deny. Modern remote sensing research has advanced from descriptive earth 'observation' through an era of data gathering and exploratory image processing to a point where the value of remote sensing data as measurements is paramount. Remotely sensed measurements can be collected in a range of data dimensions and at time and space scales which cannot be accessed in any other way. This has enabled geographical applications to proceed in terms of scientific models and process understanding. The physical link between the data and the parameters of interest to users is providing the returns that previous image processing and 'correlation' based approaches could only suggest as being possible. 'Making the transition' from close-range data that the end user can understand and needs to the recorded counts in a remotely sensed data stream is today's challenge for remote sensing.
This can be accomplished in a way that serves the needs of the established applications areas in CSIRO and the wider Australian community. The means is to thread together the internationally acclaimed skills of CSIRO researchers and provide them with the opportunity, support and facilities to jointly tackle common scientific themes and issues. The themes and issues tackled must be those which underpin the applications being pursued in the individual, industry oriented Divisions and in the growing value added industry throughout Australia.
This common base of interest includes issues relating to:
- sensor and instrument design and characteristics;
- base data processing;
- calibration and geometric/geographic modelling;
- integration with other data;
- atmospheric corrections;
- data management;
- the measurement models relating remotely sensed data and the targets of applications; and
- the most effective computational algorithms;
I believe an effective network of remote sensing scientists can be developed. It will consist of a well supported core group of scientists drawn from current CSIRO staff plus a few complementary and supporting additional staff. The core group will develop a network with remote sensing scientists in applications areas and work in projects with them to provide their areas of applications with the means to make much greater progress. This will occur as their efforts will not be dissipated in attempts to cover all the tasks and resolve all the problems arising between instrument deployment and product delivery. Such activity was identified as a common problem among the AVHRR data collectors in the Simpson Report.
The Simpson report also proposed a new focus for MDP-19 which was to develop 'Level 1.5' products that could overcome the need for applications oriented groups to solve the problems in the list above. However, I disagree with the assumption made in the Report that "science focus areas (Level 2) naturally will emerge if the Level 1.5 work is done well." The work of the Centre must be focussed on the measurement models that enable the applications. The Level 1.5 products are the raw material on which both the Centre and the applications research areas can work.
The viewpoint of the core group and associated Divisional scientists - collectively the EOC - should also not be focussed on particular instruments nor their specialised data. Rather, the EOC should emphasise applications and the measurement models that relate the targets to the appropriate range of available raw data. For example, monitoring vegetation cover has applications at many scales and finer scale data is usually needed to define cover/image relationships. If the working scale is 1:10,000 to 1:25,000 then airborne data and possibly Spot or TM data can meet the needs - but AVHRR cannot. On the other hand, at regional scale, the integration provided by AVHRR can match the scaling of the application's needs and the different platform and sensor becomes the appropriate data. The central issue, however, is that underlying all the choices are the radiative transfer in the coupled atmosphere and surface layer and its scaling behaviour.
The primary activity therefore must include flexible considerations of sensor characteristics and scale of platforms together with the way they create measurements of water covered areas, complex land surfaces and the atmosphere. The specific issues (including scale and areas of application) relating to particular instruments can be gathered productively within the measurement themes. The priorities and emphases of the groupings must be determined by the needs of the applications. The areas of application must reflect the interests of the Divisions and other clients in the Australian Earth Observation industry rather than the interests of the individual scientists.
Remote sensing and Earth Observation are not only about satellite data nor only about low resolution meteorological satellite data. The pressing needs for environmental monitoring demand data with spatial scales from 1:5,000 to 1:5,000,000 and time scales from daily to decadal. It would be a serious omission to ignore the tactical monitoring and assessment issues created by sudden or localised disturbances and natural disasters. They are as important to respond to as the longer term environmental changes being monitored by space observing systems. In addition, integrating the observations at an individual site with satellite data without concern for the intermediate scales creates a disturbing gap in calibration methodology. This gap can only be filled if there is a commitment to higher spatial, spectral and temporal resolution earth observation in the centre. It is no accident that the advanced EOS instrument - MODIS - has high spectral resolution and multi-scale data.
I believe that the EOC must also from the beginning address the issues involved with airborne (1:5,000 to 1:10,000 scale) data and the higher spatial and spectral resolution sensors they carry. This must be done within the same integrated framework of measurement models and algorithms it develops to deal with satellite data and the various sensors they carry. The scalability of processing and the interpretation algorithms is both practical and fundamental to creating successful operational and packaged links between sensor and target on any platform. Developments in instruments able to be deployed on aircraft at the present time is similar to the development of PCs in the early 80's. Very soon the range of options will be enormous and the opportunities they create for rapid tactical earth observation must not be neglected. The greatest opportunity lies in the potential for them to be calibrated and have sensitive, narrow and flexibly placed spectral bands. Their needs in this case are exactly for the science and algorithms the EOC has the capacity to provide. As an example of how measurement models can integrate with the data processing stream, I have attached a document describing some of the airborne work that has been done at CSIRO Division of water Resources. This work is currently being developed in a controlled combination of commercialisation and collaborative international research.
For the centre to be successful, there is a need to create strong threads of collaboration and interaction among the excellent scientists currently dispersed throughout CSIRO. Some should become fully part of the core EOC but others should simply be encouraged and supported to make use of the EOC and participate in joint research. This would be ensured through making significant (eg 3 or more month) visits to other laboratories and/or the EOC centres essential components of the collaboration. The EOC could consist of one primary core group in a single location plus more tempory 'satellite' groups which are developed in collaborating Divisions and centres for specific projects and studies. Exchange and significant visits between the Australian groups are alternatives to re-location. They may well provide the 'glue' the EOC needs in order to create open collaboration, trust and frequent contact between the scientists that make up the strong base of skills in fundamental remote sensing science. The budget of the EOC will need a very visible and solid allocation for this activity rather than either being restricted to the core group at the centre or dispersed in small allocations. The alternative provision of separate small 'grants' to geographically separated groups can only make the fragmentation and lack of critical mass at the separate sites worse.
The EOC should also encourage joint projects with industry. Its clients must potentially be inside and outside CSIRO. In the early 1980's, the Canadian CCRS pursued a policy of collaboration with industry in fundamental science that involved secondment of staff from industry to CCRS for extended periods as part of the R&D. This has created a strong science and research focus in the Canadian earth observation industry which has paid off considerably. Tasks that could be done by industry or that were desirable for industry to take on were always contracted out rather than done in-house - but with close collaboration and staff exchange. You cannot assist industry by competing with it. I believe this is the cooperative model the EOC should also adopt. In doing so, the Centre will also need to address the concerns of industry for standards, economic evaluation of alternative data sources, information throughput and the cost-effectiveness of different levels of value adding. It is through industry contact that the Centre can contribute to these important analyses.
It is no accident that centres such as the EOC have emerged in recent years in other countries and are developing a strong network between their scientists. The issues that must be tackled to realise the opportunities of an EOC are international by nature. They can only be successfully resolved by very close and collaborative contact between researchers at similar centres in different countries. In particular, links must be made with the US, Canada, Europe, China, Japan and the ASEAN members. Without such strong international collaboration the EOC may achieve less than is being achieved now by the dispersed but internationally active groups. The EOC needs to put special emphasis on promoting and enabling contact, collaboration and joint publication between Australian and other scientists working on remote sensing science throughout the world. Bringing scientists to Australia and sending Australians overseas are fundamental activities once the internal contacts have been established. Participation in international projects and forums will pay-off significantly in this area and must be encouraged. It should be emphasised, however, that this cannot be achieved only through participation in international committees but by international collaboration.
The EOC will succeed if it creates an environment in which the solution to the 'Prisoner's Dilemma' is carefully encouraged to be cooperation rather than to take what is offered but give as little as possible. As a person who has worked for many years to see remote sensing applied I believe the EOC can enhance the flow of applications if its focus is well directed. However, to succeed in the face of realities, I believe it will also need to have strong support from Executive level down and have a high degree of independence, discretion and support. The Advisory Committee (for example) should not run the EOC but be like the Board of a Company with the Science Program Leader as MD. The 'MD' also needs the capacity to make decisions and make them stick. There may come times when the success or total failure of the EOC depends on the leverage that is provided to the EOC. Speaking softly but carrying a big stick was never more important. The next year will be an exciting one for all concerned with the venture but where it all ends will depend on the decisions made in the very near future.
Attachment: Integrated Airborne Sensing at CSIRO Division of Water Resources
Use of this web site and information available from it is subject to our
Legal Notice and Disclaimer