1 Centre for Resource and Environmental Studies, The Australian National University
2 Department of Forestry, The Australian National University
Phone: (02) 6125 8129, Fax: (02) 6125 0757
Increased responsibility for amelioration of land degradation and associated planning has shifted to community based groups such as Landcare. For these groups the development of catchment and regional plans to effectively allocate resources are constrained by the availability of information at appropriate scales. Geographic Information Systems (GIS) and associated remotely sensed data are seen as useful in improving delivery of this information.
This paper describes two separate case studies of land cover mapping that examine the role of GIS for Landcare groups. The case studies contrast in terms of scale, 10’s to 1000’s km2; organisational structures, local to regional; physical environments, grazing and remnant native vegetation to extensive cropping; and objectives, pasture to remnant vegetation mapping. The paper evaluates the approaches and makes recommendations for the use of GIS and remote sensing by regional and catchment land management organisations. In particular the usual plethora of claims made by technicians in GIS and remote sensing are considered in the context of organisations such as local and regional Landcare.
GIS and remote sensing are effective tools in Landcare, but consideration must be given to many factors that include the structure, capabilities and information needs of the organisation, together with the potential cost and benefits.
Land degradation processes and problems are often not constrained by property boundaries. Generally, catchment or regional scale approaches are required to effectively deal with these problems. As traditional extension services decline, due to reduced government funding, Landcare organisations are taking increased responsibility for catchment and regional planning. For these Landcare groups a major constraint to the development of effective management plans at catchment scales is a lack of appropriate information. Geographic Information Systems (GIS) and data sourced from remote sensing are seen by these organisations as potentially attractive means of providing this information.
Very few studies, other than at broad regional and State levels, have been undertaken to investigate the use of GIS by Landcare. Broad overview studies tend to produce evaluations at scales removed from the context of the users of geographical information. Hence, there is a need for smaller, case study based research (Ewing 1997). The aim of this paper is to investigate the role of remote sensing and GIS in provision of information to Landcare groups. Two contrasting case studies of resource assessment and planning within separate Landcare organisations are examined.
A short review of the information needs of Landcare groups is presented first. This is followed by a description of each of the case studies. A discussion of the lessons learnt from the case studies is then made with an alternative structure proposed to better deliver spatial products to Landcare organisations.
Landcare and Information Needs
Landcare is a collection of community-based organisations working towards sustainable land management. Landcare has evolved as a ‘grass-roots’ movement in response to land degradation problems at the local scale (Baker 1997). However one of the primary drivers for the ongoing development of Landcare is that actions to ameliorate and prevent land degradation are necessary over scales larger than the individual farm due to the often spatially separate causes and effects of land degradation. Community based Landcare groups started to form in the early to mid-1980’s (Baker 1997). It is estimated that there are now in excess of 4200 community based Landcare groups Australia wide (Quealy 1998). Approximately 34% of farms have a property representative who is a Landcare member (Mues et al. 1998).
Landcare is a relatively new and evolving organisation with a diverse and dynamic structure. These structures attempt to deliver a coordinated approach to addressing environmental problems. At the core of the Australian Landcare network is the local Landcare group (Quealy 1998). Typical local Landcare groups consist of 20-30 individuals who are directly concerned with the implementation of on-ground works and the understanding of local conditions.
Many local Landcare groups find that they face similar degradation problems as those of their neighbours. The problems may flow onto, or be caused by areas represented by neighbouring groups. This results in a need to provide a coordinated regional approach and has led to the formation of regional Landcare groups. Regional organisations tend to be grouped in catchment areas and assume responsibility for the coordination of resource management within their catchments. Regional groups are generally successful in attracting funding for full-time coordinators to facilitate the activities of both the regional organisation and assist individual Landcare groups. State and national Landcare organisations exist, however, these are not considered in this paper.
To make good decisions Landcare organisations must have appropriate information about their local area and a context within the catchment and region (Hill et al. 1996). Improving the availability and distribution of this information is critical for the widespread adoption of sustainable land use practices (Commonwealth of Australia 1991). According to Hill and coworkers (1996) to fulfil the information needs of Landcare groups the following is required:
- access to mapping to address land management issues at a farm, Landcare and catchment scales, which may include data about land cover, topography, soils, geology, hydrology and cadastre;
- ability to collate and integrate community monitoring information and to analyse, interpret and distribute the results; and
- access to appropriate advice and other tools such as decision support systems where necessary.
There has been criticism that the information available to Landcare is not good enough (Bailey 1997). Problems exist for many Landcare organisations in being able to access basic biophysical data that results in difficulty in making fully informed decisions (Carr 1997). Similarly, frustration is caused to individual Landcare group members who are unable to gain a sense of how their localised action and efforts contribute to the ‘big picture’ (Carr 1997). Poor coordination between land management organisations, and data ownership and access issues are part of the problem, however there is also a simple lack of data (Hill et al. 1996).
An estimated 56% of farmers use Landcare as a source of farm management information which is significantly higher that the total proportion of Landcare membership (Mues et al. 1998). The information that is sought by farmers from Landcare is very often to address geographically specific problems. District, catchment or regional planning is taken into account by 45% of farm planning, and a little under half of farmers who have prepared farm plans take into account the property layout or farm plan of a neighbour (Mues et al. 1998).
Landcare groups are attracted to the use of GIS as an efficient, cost effective and accurate means of providing information (McGowen 1996). GIS allow Landcare groups to interactivity plan, evaluate and measure the success of projects, identify problems and opportunities for on-ground work and use available information more efficiently. Landcare organisations should consider the use of remote sensing and GIS when the land management information that they seek is required to have extensive coverage, internal consistency and the ability to be reproduced.
Unfortunately the experiences of a number of groups in the use of GIS and remote sensing has not been encouraging. These experiences provided the impetus for this research. Problems can be attributed to a number of factors:
- a dichotomy between the expectations of Landcare groups and the actual ability of the data provider to supply information given commercial constraints on time, labour, software and data;
- the raising of unrealistic data and skill limitations and expectations of the results and GIS capabilities, elevated as part of the hard sell of consultants;
- high costs involved in the setting up of GIS and the staffing of projects; and
- limited training, experience and understanding of GIS and remote sensing technologies by individuals within Landcare groups.
Remote sensing and GIS based resource assessment case studies were undertaken for the Ballalaba-Krawarree Landcare Group (BKLG) and the Boorowa Regional Catchment Committee (BRCC). These two case studies were selected to contrast organisational structures, farming systems and physical environments. Both case studies were undertaken to provide information on the land cover of the catchments: the Boorowa case study to investigate pasture types and distribution; and the Ballalaba-Krawarree case study to identify and classify areas of remnant forest and woodland vegetation.
Brief detail of the methods and results of the case study are presented in the following sections. Only a brief presentation of the methods and results of the case studies is made to allow for the interactions between the information users and providers to be more fully described.
Case Study 1: Pasture Mapping, Boorowa River Catchment
The BRCC, a regional Landcare organisation, required information on the extent and distribution of pasture types across the Boorowa River and adjacent Hovell’s Creek catchments. The information was required to target on-ground perennial pasture establishment, aimed at lowering watertables in an attempt to reduce the increasing problem of dryland salinity in the catchment.
A maximum likelihood supervised classification of multi-temporal Landsat TM imagery was to used differentiate between perennial, annual and native pasture types and other land cover categories – cropping, lucerne, woodland and water. Land cover types rarely have unique and unchanging spectral signatures, but differences in growth rates, maturity and structure at certain times of the year provide reflectance characteristics that aid in discrimination. Thus imagery was selected to display a difference between the spectral response of perennial pastures to contrast with annual and native pasture types. An image from October 1997 – where all pasture is actively growing and an image from December 1998 – a dry period, where only perennial pastures were still growing, were selected for analysis.
Training sites were selected with the assistance of the Landcare coordinator and cooperation of landholders. Training sites were located with a differential GPS. An analysis to ensure that training sites were spread across a range of environmental variables – as a surrogate for background spectral response, was made. Independent accuracy assessment sites were also collected and located with a differential GPS.
The classification was successful yielding an overall accuracy of 62% from independent accuracy assessment sites. Few similar projects in the mapping of pasture have documented their levels of accuracy. Without comparison, and given the similarity in the spectral response of the pastures, this is considered a good result.
The BRCC does have limited GIS facilities – using the ARC View GIS software package. This allows for the interactive display of the results of the case study. With this system results can be more effectively distributed to the local Landcare groups of the area and the data updated as new information is gathered. Informal training sessions in the use of GIS software were provided to the Landcare coordinator and selected members of the regional committee as part of the project. Another function of the project was to bring together data sets available to the group under a common map projection and standard. The training provided to the Landcare coordinator and the bringing together of available data sets were outcomes that were valued as importantly as the land cover mapping, by the group.
The results of the Boorowa case study demonstrate that remote sensing and GIS are useful in determining the type and extent of perennial pastures and other land covers. The information derived from the investigation allows the BRCC and the local Landcare groups to strategically target on-ground work aimed at reducing dryland salinity. The structure of the BRCC organisation and their access to a GIS enable the committee and individual members to use and distribute the results of the study effectively.
Case Study 2: Land Cover Mapping, Upper Shoalhaven River Catchment
The Shoalhaven case study was undertaken for the BKLG. The group is a local, community-based group with a particular interest in the conservation of remnant vegetation. The aim of the case study was to identify the floristic composition of the remnant forest and woodland vegetation areas in the upper Shoalhaven River catchment. The information derived from the work was to be used by the group to develop vegetation management plans aimed at retaining and restoring the remnant vegetation of the catchment.
The project was performed after the failure of a consultant employed by the group to produce results to the standards set by the group. A particular discrepancy between the group’s expectation of a floristic survey and the provision of a vegetation structure map by the consultant proved insurmountable.
The land cover of the case study area was determined by a supervised classification of both spectral and environmental data. The classification was undertaken in two stages. The first stage was to classify the spectrally distinct land cover categories by supervised classification of combined SPOT XS and SPOT panchromatic imagery. The second stage was to make a classification using combination of spectral and environmental attributes (altitude, slope and aspect) to distinguish the land cover categories of the native forest areas that were effectively indistinguishable using spectral characteristics of the land cover alone. Minimum distance supervised classification techniques were used.
The procedure produced excellent results. From an accuracy assessment conducted using independent reference data an overall classification accuracy of 66% was achieved.
The major limitation of the Shoalhaven case study was the classification method used. Research is required to provide better techniques for the production of thematic classifications from combinations of spectral and environmental data. Whilst many methods are presented in the literature, they are generally experimental techniques that require extensive computer programming and lie outside the resources of Landcare. Other environmental input data – climate surfaces, position on slope measures and soils data, at an applicable scale, could all improve classifications of this nature.
Problems were experienced in the case study because of the unknown origin of data sets. Landcare groups (and consultants) must ensure that the quality and origins of all data sets are known (and recorded) to provide firm foundations on which to base their remote sensing and GIS activities.
The information provided from a combined remote sensing and GIS investigation in the Shoalhaven case study provided both the type and distribution of remnant forest and woodland vegetation. The case study demonstrated that additional environmental data aids the successful discrimination of native forest land cover categories. The BKLG is constrained in its ability to interactively use the information derived from the investigation, as the group does not have access to GIS software and therefore no training could be provided.
The background to this paper detailed the increasing advisory role of Landcare and the problems these organisations face in accessing basic biophysical data. These pressures will undoubtedly lead to a future increase in the role that remote sensing and GIS play in providing land management information for Landcare. This information, needed at a scale and precision which allows for the development of management recommendations for use by regional and local community Landcare groups, is necessary to efficiently target their limited resources. In the case studies, remote sensing and GIS were shown to be useful tools for the provision of this spatial information. However, as can be seen in the case studies, the expectations of Landcare groups are not always met by applications of remote sensing.
The efficient incorporation of remote sensing and GIS into Landcare requires careful consideration of the organisational structures, the information needs of small land management organisations, the technical capability available to Landcare organisations and a consideration of the information that can be provided by remote sensing and GIS.
Of the case studies, the Boorowa example provides the best level of the organisational structure for the effective incorporation of remote sensing and GIS. In the Boorowa case study the information needs of individual Landcare groups were assessed by the regional Landcare coordinator directly and the project tailored to meet these requirements. The general familiarity with computing generally, and GIS and remote sensing specifically, that the coordinator already possessed made the project more realistic, manageable and successful. Further, the availability of interactive operating GIS software in-house allows for more effective communication and dissemination of results to users. More complex tasks such as the case study example are best undertaken outside of the group and the data added to the existing system. Regional groups have the capability to provide good supervision and support of projects due to the generally greater knowledge of remote sensing and GIS applications among Landcare coordinators at this level.
To move down the organisational structure and to base remote sensing and GIS activities at the community Landcare level leads to a number of problems in specifying project requirements and reduces the ability of the users to effectively supervise projects that are necessarily undertaken outside of the organisation. Evidence of this is the failure of a consultant employed by the BKLG to deliver the results expected. Such problems may have been avoided if the group were in a position to provide better project specification and supervision. The group did not fully understand the capabilities of remote sensing; they had expectations, that remotely sensed land cover classifications could be made at a species level and that the imagery would enable landholders to plan the management of small areas of their individual properties. They were partially correct on the first and clearly mislead on the latter. The voluntary labour that makes up the community Landcare movement often cannot be expected to adequately supervise potentially expensive, complicated and detailed scientific projects. At the local level of Landcare there may not be sufficient resources to commit to setting up and maintaining a GIS for the processing, display and query of geographic data.
Whilst not explored in this research to move up the organisational structure to State and National levels of Landcare would result in GIS’s that are too far removed from the needs of the users of the information to be effective.
There is a need for a better way to integrate and distribute information from remote sensing and GIS applications. In too many instances the information that is delivered provides attractive, yet somewhat expensive, wallpaper! Results from GIS applications are more valuable when they can be interactively displayed and used. This enables Landcare groups to investigate where on-ground work should be located in catchment and regional frameworks. All new projects should be evaluated with reference to the spatial information that is available to the group. Any new data produced outside of the Landcare group is required to be delivered in a form (and format) that is easily interpreted by the expertise available to the Landcare group.
The technical limitations of GIS are commonly problems in the communication between users and the technical providers of information. Evidence of this is the success of the Boorowa case study where communication with the BRCC was effective and results obtained were good. This is in contrast to the perceived poor results provided by the private consultant employed by the BKLG where communication was generally insufficient.
Separate groups have developed within the GIS community: those that process data and those that are responsible for the interpretation of that data, and transfer of the results to user groups (Vickery 1994). Efficient communication is required between the people that require the information and the people who are responsible for data delivery. Whilst it is undesirable for the two distinct groups to emerge, Landcare (and reasonably technically educated Landcare coordinators) may provide a better way of interfacing between the remote sensing community and the end users of this information. Educated Landcare coordinators are generally the key to that efficient transfer. To understand the sorts of spatial data manipulation operations of importance in GIS one must take the view of the resource manager rather than that of the data analyst to provide management information (Richards 1993). In the case studies regular consultation and presentation of results achieved this with the two groups involved.
A fundamental requirement for the acceptance of remote sensing is that the resultant products be reliable. Valid assessments of accuracy are necessary for evaluating the quality of the data prior to making management decisions; these assessments need to be planned as an integral part of each project.
The interactive capabilities of GIS, unlike traditional forms of resource assessment, allow Landcare groups to progressively improve on the accuracy of resource assessments by correcting existing data sets as new information, such as additional field investigation, is undertaken.
Despite considerable investment in the development of agricultural and resource management applications of remote sensing and GIS, there are surprisingly few economic cost-benefit analyses documented (Watson et al. 1995). It is difficult to quantify what the full range of benefits and costs of applications of remote sensing are, as many are of a non-market nature (Watson et al. 1995).
Developments that are impacting on the present, and have the potential to change the future role of remote sensing and GIS in Landcare include:
- An increase in the quality (spectral and spatial) of remotely sensed data foreshadowed by many including Lillesand and Kiefer (1994). The improvements will undoubtedly result in better classifications of land cover and other resource assessment and will broaden the potential use of remote sensing. Improved spectral discrimination, particularly in the Boorowa case study, where the classification was dependent on discrimination of land cover with very similar spectral response, will allow for greater accuracy to be achieved.
- Increased computing power and availability of personal computers and decreasing cost is expected to continue into the future. The implication of this for Landcare is that there will be a reduction in the set-up costs of GIS for Landcare organisations and an increase in the subsequent use and acceptance of the results of remote sensing.
- Increased ease of use and lowering of cost of image processing and geographic information systems. The effect of these changes will reduce the levels of training and experience required for successful use of these systems thus enabling Landcare organisations to make better use of GIS.
- The potential dissemination of geographic data on-line is an interesting development in information availability for Landcare. Australian farmers have shown a rapid uptake of Internet technologies with 14% of farmers already on-line (Simpson 1999). With only 2% of farmers according to Simpson (1999) regarding the Internet as a poor farm business investment, this figure is increasing. Advantages of placing the results of Landcare research on the Internet are immediacy, low cost and access to a broad audience. If farmers continue their uptake of Internet technologies at their current rate, research organisations and Landcare will be required to provide interactive use of geographic data and publishing on the Internet. The Internet will improve transfer and dissemination of land management information. The Internet enables many usual functions of GIS to be performed remotely including the on-line query of the result of remote sensing and GIS research. For example publication of the results of the Boorowa case study by the BRCC would allow access to this data by a large range of both Landcare and non-Landcare farmers, both within and outside the catchment. From the Landcare perspective the future publishing of geographic information on the Internet is not without emerging difficulties, including quality, copyright, rural access, legal and ethical issues (Crowley and Steinke 1996; Peter 1996; Stewart et al. 1997; Slonecker et al. 1998; Herd 1999).
- A shift overseas toward provision of remote sensing products by the commercial sector identified by (Slonecker et al. 1998). Presently, much of Australian remote sensing activities are focused on research and development activities, primarily in the government sector (Watson et al. 1995). The implications for Landcare if this trend is mirrored in Australia are unclear. One implication may be an increased reliance on commercial consulting services and hence a requirement for detailed project specifications to be provided by Landcare groups.
These points together will make GIS and remote sensing much more accessible to Landcare groups. No longer will GIS be, to quote White (1996), ‘the preserve of backroom high tech whiz kids’. These advances may allow individual land managers and community based Landcare groups such as the BKLG to use these tools for their independent investigation. However, at least in the short term, there will still be requirements for the processing of geographic data outside of the Landcare organisation.
The following model is proposed as an optimum way to incorporate remote sensing and GIS into the Landcare structure.
It is proposed that remote sensing and GIS activities be undertaken at the regional Landcare level. GIS systems are required at this level for display and interpretation of geographic data for the effective dissemination of results. The group as part of this system should record all geographic data. Detailed projects should be undertaken outside of the Landcare organisation. Groups should seek independent technical advice when contemplating new remote sensing and GIS activities and projects. This advice from GIS and remote sensing experts will avoid problems in project specification, such as those experienced by the BKLG. Sources for such advice may include government departments and educational institutions along with advice from within the Landcare movement. Individuals and community Landcare groups need to have access to the information derived from remote sensing and GIS investigations tailored to their needs.
An examination of the role that remote sensing and GIS can play in two detailed case studies of land cover mapping has been extended to see where and how remote sensing and GIS should be incorporated into the Landcare organisation for the greatest benefit. It is clear that for GIS and remote sensing to be effective tools in Landcare, consideration must be given to many factors that include the structure, capabilities and information needs of the organisation, together with the potential cost and benefits of GIS and remote sensing.
From examining the case studies it can be concluded that remote sensing and GIS activities should be primarily incorporated at the regional level of Landcare organisations. A number of factors were considered in arriving at this conclusion. The Landcare organisation requires information predominantly at catchment to regional scales and the characteristics of remotely sensed data, particularly an extensive coverage, make such a scale suitable. At the regional level there are often the resources and skills within or available to the group, to adequately undertake or supervise remote sensing and GIS activities. Similarly, the regional level of the Landcare is well suited to applications of remote sensing as government funding favour projects located within regional plans of land degradation amelioration.
Advances in the quality of remotely sensed data and interpretation techniques will improve results and broaden the potential applications of remote sensing. Decreasing cost of computer hardware and software, increasing ease of use of GIS software and Internet based dissemination offers continued opportunities for Landcare organisations to use these tools. Present trends in rural information needs and the decline in traditional extension activities indicate an increase in the future use of remote sensing and GIS by Landcare organisations. Landcare groups need to be aware of potential developments in remote sensing and GIS to take full advantage of these tools.
There is a need, as always, to further document the remote sensing and GIS activities of Landcare organisations. Further investigation into economic cost-benefit analysis of the use of remote sensing generally, but particularly for Landcare, is required.
In both case studies subsequent investigations are needed to determine how the resource assessments have aided in the preparation of management plans for the groups involved, and if necessary, provide recommendations for future commissions.
Bailey, M. (1997). 'Landcare: myth or reality?', in Critical Landcare, eds S. Lockie and F. Vanclay, Centre for Rural Social Research, Charles Sturt University, Wagga Wagga.
Baker, R. (1997). 'Landcare: Policy, practice and partnerships', Australian Geographical Studies, vol. 35, no. 1, pp. 61-73.
Carr, A. (1997). 'Innovation of diffusion: Landcare and information exchange', in Critical Landcare, eds S. Lockie and F. Vanclay, Centre for Rural Social Research, Charles Sturt University, Wagga Wagga.
Commonwealth of Australia (1991). Decade of Landcare Plan, Australian Government Publishing Service, Canberra.
Crowley, T. and Steinke, A.P. (1996). 'A web of information - The future of remote sensing and Geographical Information Systems', Proceedings of the 8th Australasian Remote Sensing Conference, Canberra.
Ewing, S. (1997). ''Small is beautiful': the place of the case study in Landcare evaluation', in Critical Landcare, eds S. Lockie and F. Vanclay, Centre for Rural Social Research, Charles Sturt University, Wagga Wagga.
Herd, A. (1999). 'Using the Internet - The legal situation', Research Results on the Internet: Are Farmers' Needs in Conflict with Researchers' Interests?, Australian Institute of Agricultural Science and Technology Seminar, ACT and Southern NSW Branch, Canberra.
Hill, S., McAllister, A. and Jamieson, P. (1996). 'Natural resource information for communities', in Landcare Languages: A Communication Manual for Landcare, ed V. A. Brown, Department of Primary Industries and Energy, Canberra.
Lillesand, T.M. and Kiefer, R.W. (1994). Remote Sensing and Image Interpretation, 3rd ed, Wiley and Sons, New York.
McGowen, I. (1996). 'Remote sensing and GIS for better land management on the Jemalong/Wyldes Plains', Proceedings of the 8th Australasian Remote Sensing Conference, Canberra.
Mues, C., Chapman, L. and Hilst, R.V. (1998). Promoting Improved Land Management Practices on Australian Farms: A Survey of Landcare and Land Management Related Programs, ABARE, Canberra.
Peter, I. (1996). 'Landcare and information technology', in Landcare Languages: A Communication Manual for Landcare, ed V. A. Brown, Department of Primary Industries and Energy, Canberra.
Quealy, J. (1998). 'Landcare structure in Australia', in Landcare Australia, Ninth Annual Report and Yearbook 1998, J. Blackadder Executive Media Pty Ltd.
Richards, J.A. (1993). Remote Sensing Digital Image Analysis: An Introduction, 2nd ed, Springer-Verlag.
Simpson, R. (1999). 'The Internet vision for the future: Rural communities and the Internet', Research Results on the Internet: Are Farmers' Needs in Conflict with Researchers' Interests?, Australian Institute of Agricultural Science and Technology Seminar, ACT and Southern NSW Branch, Canberra.
Slonecker, E.T., Shaw, D.M. and Lillesand, T.M. (1998). 'Emerging legal and ethical issues in advanced remote sensing technology', Photogrammetric Engineering and Remote Sensing, vol. 64, no. 6, pp. 589-595.
Stewart, K., Cho, G. and Clark, E. (1997). 'Geographical Information Systems and legal liability', Journal of Law and Information Science, vol. 8, no. 1, pp. 84-112.
Vickery, P.J. (1994). 'Pasture fertilizer management by satellite remote sensing: A technology transfer conundrum', NSW Remote Sensing Committee Newsletter, vol. 5, no. 2, pp. 32-33.
Watson, B., Anandajayasekeram, P., Miller, E., Allen, R., Curtotti, R. and Johnston, R. (1995) Remote Sensing - Use in Agricultural and Resource Management: A Conceptual Framework for Economic Analysis, ABARE Research Report 95.5, Canberra.
White, T.D. (1996). 'Land literacy', in Landcare Languages: A Communication Manual for Landcare, ed V. A. Brown, Department of Primary Industries and Energy, Canberra.