What Is Meant by Land Use Change?
Isaak S. Zonneveld
Because of the vertical and horizontal heterogeneity of landscapes, researchers from many disciplines use land survey data. Zoologically oriented landscape ecologists study the effects of horizontal heterogeneity on animal populations (Merriam, 1984; Forman, 1982). Similarly, the data can be used to help answer a key question for humankind: Is the survival of groups of people essentially dependent on landscape heterogeneity? Agriculture and other human activities imply it is.
Landscape ecology is concerned with the study of land or landscape, its form, function, and genesis (change). It looks at the factors interacting at the earth's surface, including the physical, biological, and noospherical actions originated by humans. These factors form three-dimensional phenomena that can be seen as horizontal patterns of related elements (units of land) and as vertical patterns of land attributes, such as climate, rock, soil, water, and vegetation. The heterogeneity of these patterns is the main focus of landscape ecology.
A landscape is viewed as a holistic entity that is composed of a variety of relationships in a relatively steady state. The maintenance of a steady state is called homeostasis, which refers to the set of positive and negative feedback factors that keep the system in a dynamic equilibrium. The steady state may evolve into another steady state over time, but it is protected from strong fluctuations by feedback factors (homeorhesis).
Considering a landscape holistically implies the study of it as a whole rather than focusing only on the functioning of its parts. Such an approach allows the reduction of analytic observations needed to study very complex structures. It rejects the study of separate pieces without connecting them with each other. Because landscape ecology entails the study of the landscape and its many functions in their entirety, it is necessarily multidisciplinary.
The knowledge gained about the relationships among the parts of a landscape and its function as a system can be used as a basis for planning and managing land use. The policy of humans should be to maintain environmental sustainability through stable landscape configurations. Landscapes can be configured to attain human objectives, such as food production. Each configuration affects ecological integrity differently. The character as well as the configuration of the landscape elements determine to a high degree the stability (in the sense of persistence) of the landscape in relation to disturbances to its steady state (Forman, 1989, and private correspondence).
Land use is the varying activities executed by humans to exploit the landscape, such as hunting or ploughing. The land use pattern primarily determines the landscape pattern in areas where land use is intensifying. Therefore, land use patterns may be essential factors in determining landscape stability and should have the attention of land use planners.
The simplest example of manipulating landscape stability by pattern is the contour arrangement of land parcels to prevent soil erosion. Contour ploughing is at the smallest scale of landscape management. At larger scales it is the conservation of permanent vegetation (usually forest) in the water catchment areas of river basins.
More complex is the concept of ecological infrastructure or the importance of connecting landscapes to maintain biodiversity. This concept accepts the necessity of exchange among biological populations. A certain separation is necessary for individual development (evolution). But if the separation becomes complete isolation, because of the fragmentation of landscape due to highly intensive agriculture, it could lead to genetic impoverishment and the extinction of species.
Research on the connectivity of landscapes for groups of species is developing as data on area, distances between patches, stepping stones, and the character of corridors are collected and disseminated. For local planning needs, use can already be made of such knowledge. On a more global scale, the research is less developed. In The Netherlands an attempt has been made to include ecological infrastructure in state land use planning. Implementation is in its early stages. The first steps (on paper) for Europe are in preparation (Benneth, 1991).
More common practices are unplanned extensions of agricultural land that violate the connectivity of landscapes in originally stable ecosystems.
Examples include creating large agricultural or hydrological projects across well-known migratory routes of animals, such as was done to the elephants in Sri Lanka and other animals in Sudan (Cox, 1988). Better examples are the intensification of agricultural production on existing arable land of good quality, optimally using biologically diverse crops. For a real understanding, however, additional research effort and fieldwork will be required.
MEASUREMENT OF LAND USE CHANGE BY REMOTE SENSING
The study of landscapes entails the measurement of land use. The result of human activity is partially seen in the changes in land cover, i.e., the conversion of natural forest to farmland and beaches to urban centers. Measuring these changes involves measuring land cover at different time intervals. Human activity can be measured by asking hunters how many of them are hunting what, when, where, and with what intensity, or by asking farmers what kinds of crops they are cultivating when, where, and how.
Measuring can also be done by assessing the results of human activity, the land cover. This can be observed visually, directly with the eye or by photographs or other remote-sensing methods. Ideally, one would use a combination of measuring the activity and the land cover for each land area of interest.
Measuring land cover is the most suitable type of measurement for global observations. Measuring human activity generally requires more time and energy because visits to the area to be measured and discussions with the people living there are necessary. In administratively and technically advanced countries, statistics are available as well as a series of topographical maps done over time with which to compare the most recent measurements. In less advanced countries, statistics on land use tend to be less accurate and up to date, and topographical mapping is less frequent and of insufficient detail. In these countries, measurement by remote sensing in combination with some fieldwork, to establish the "ground truth," is the only way to come to reasonable results.
In developing countries, assessment of land use over large areas is often difficult without remote sensing. In combination with statistical data, such as those from an agricultural census, and ground truth, remote sensing meets basic land use measurement needs. However, the accuracy of land use classification depends on the scale of the observations, which may be too small (i.e., not much detail) for some planning needs.
The most common means of remote sensing is the use of various wavelengths of electromagnetic radiation. This radiation is reflected from objects on the earth's surface and can be recorded by photography or elec-
tronic devices. In all cases, only a limited part of the light spectrum is used.
Infrared radiation is a very useful type of radiation to record because it represents a considerable percentage of the sun's energy that reaches the earth's surface. It is strongly absorbed by water and wet soils and is not used by plants for photosynthesis, so that the energy is reflected by plant tissues to a large extent.
Observations by remote sensing are predominantly made from aircraft for intermediate-scale images. Satellites are used for small-scale images, and very detailed (large-scale) images are taken from fixed platforms, balloons, or small aircraft.
The choice between the various scales and means for remote sensing depends on the nature of the land area, the purpose for the resulting information, and the general economic and political situation (which often determines the availability of data for nonmilitary use). The most universal means for reconnaissance mapping is ordinary black-and-white photography taken at the scale 1:40,000 with a super-wide-angle lens from conventional aircraft.
Interpretation of Remote Sensing Data
The results of the photography and scanning are images that need to be interpreted to transfer the data into usable information. In some cases, special image processing will be done at the same time.
The patterns on a remote-sensing image consist of patches differing in color or gray hue. A most important feature for landscape analysis and classification is the configuration of the pattern elements and how they relate to various types of landscape mosaics. There are random mosaics, regular mosaics like checkerboard ones (sometimes associated with farming), or dot mosaics. These mosaics imply certain ecological infrastructures (Zonneveld, 1988; Forman, 1989; Forman and Godron, 1986).
Essential in photo interpretation is the vertical dimension of the landscape, especially on large or intermediate scales. The large variations in the vertical dimension indicate variations in land relief (e.g., as mountains and oceans), and the smaller variations show the height of ground vegetation. The most important disadvantage of satellite imagery is that vegetation height and low-level land relief are impossible to detect due to the very high altitude from which the images are taken.
Once the various elements on the remote-sensing image have been differentiated into single homogeneous units or, depending on the scale of the image, complexes, then ground truth knowledge is gathered. Stratified sampling is carried out based on the interpretation map of the image, preferably in relatively small representative areas. In this process, various types of land
cover on the ground are compared with the image. Surveys, using questionnaires, of functional land use can also be carried out at this stage. Observations from small aircraft can be used to replace ground checks of cover type in areas that are inaccessible. This ground truth is used to more fully interpret the remote-sensing images already gathered (Küchler and Zonneveld, 1988).
Since the early days of remote sensing, people have tried to automatize interpretation. The practical possibilities for automation are less than the layman would expect. Nevertheless, the development of electronically recorded remote-sensing images does allow for certain automated classification procedures.
The best opportunities are in areas with large agricultural fields that are easily distinguishable from adjacent areas of natural vegetation. Computer-assisted interpretation is most effective with large homogeneous map elements, such as the U.S. cornfields. Care must be taken in such interpretation because slight differences in superficial structure (roughness, etc.) or temporal changes due to wind or small differences in the growing stages of crops may translate into large differences on the image. In contrast, different crops may appear as the same image. It is particularly difficult to automatize vertical features of landscapes (Kannegieter, 1988; Sombat and van der Zee, 1987).
Use of Remote-Sensing Data
The data collected from land use surveys is used for:
governmental planning purposes of various kinds,
commercial assessments of the supply of a land product,
determining target areas for humanitarian relief efforts, and
studying environmental degradation and its consequences.
The data should be able to provide answers to the following questions:
How is the land used?
How has land use changed?
What and how much is produced by the land? And
Who uses the land?
The main constraints in developing countries to utilizing the results of land use surveys are the lack or inaccuracy of statistical (agricultural, cadastral,
topographical) data and the irregularity of the land in space and time (i.e., shifting cultivation). Quantitative data on agricultural production are also much more difficult to infer from aerial photographs and satellite imagery than they are in countries like the United States or the former USSR, where large parcels of land with certain spectral signatures allow straightforward calculation. However, in certain regions with characteristic complex land use patterns and distinct growing seasons, it is possible to calculate general data about production potential (Groten, 1991).
An enormous constraint to collecting usable data in many developing countries is the restriction in use of aerial photography due to security reasons. This restriction often necessitates the use of satellite imagery, which, as discussed above, is less detailed than aerial photography.
However, satellite imagery captures large areas in single photos and can be used to distinguish general land use types. Marked differences between official statistics and reality can often be assessed. It is possible, particularly in the dry season, to distinguish between irrigated and nonirrigated agriculture, such as in Sri Lanka and Tunisia. Deforestation can also often be assessed, although it is often difficult to distinguish primary from secondary forests. If seasonal differences exist, however, multitemporal images may provide discrimination possibilities.
In humid tropical zones, it is difficult to obtain sufficiently cloud-free satellite images (Groten, 1991; van der Zee and Cox, 1988). Side Looking Airbourne Radar, which can "look through the clouds," can be used in these areas, but is more expensive (Sicco Smit, 1988). The author's experience in Amazonia revealed that shifting cultivation and coffee plantations can easily be detected on SLAR images.
In most cases, images collected over time are used for monitoring changes in a land area or for "watching in order to warn." For example, such monitoring can point to widespread crop failures. To study seasonal differences, monthly or shorter temporal resolution is required (Groten, 1991). For long-term planning and monitoring, yearly or longer period image collecting is sufficient.
In principle, any land survey method can be used to detect change over time. Compared to a single ad hoc survey, however, costs and compatibility with automation systems are a more important consideration. For this reason, sequential satellite observation has great advantages. It is clear that modern geographic information systems improve considerably the processing and retrieving of sequential survey results.
I am grateful to my ITC colleagues, Dr. S. Groten, Professor N. Mulder, and Dr. D. van der Zee, for their suggestions about and corrections of the content.
Benneth, G. 1991 Towards a European Ecological Network. The Hague, The Netherlands: Ministry of Agriculture, Nature Management and Fisheries.
Cox, J.A. 1988 Remote sensing and land evaluation for planning elephant corridors in Sri Lanka. ITC Journal 2:172–177.
Forman, R.T.T. 1982 Interaction among landscape elements: a core of landscape ecology. Pp. 35–48 in S.P. Tjallingii and A.A. de Veer, eds., Perspectives in Landscape Ecology. Proceedings of the International Congress of the Netherlands Society of Landscape Ecology. Wageningen, The Netherlands: Centrum voor Landbouwpublikaties en landbouwdocumentatie Wageningen.
1989 Ecologically sustainable landscapes: the role of spatial configuration. Pp. 261–278 in I.S. Zonneveld and R.T.T Forman, eds., Changing Landscape: An Ecological Perspective. New York: Springer-Verlag.
Forman, R.T.T., and M. Godron 1986 Landscape Ecology. New York: John Wiley & Sons.
Groten, S.M.E. 1991 Satelliten-Montoring von Agrar-Okosystemen im Sahel, (International Institute of Aerial Survey and Earth Science, Enschede, The Netherlands). Inaugural Dissertation, Westfalischen Wiljhelmsuniversitat, Munster, Germany.
Kannegieter, A. 1988 Mapping land-use. Pp. 335–374 in A.W. Küchler and I.S. Zonneveld, eds., Vegetation Mapping, Handbook of Vegetation Science. Dordrecht, The Netherlands: Kluwer Academic Publishers.
Küchler, A.W., and I.S. Zonneveld, eds. 1988 Vegetation Mapping, Handbook of Vegetation Science. Dordrecht, The Netherlands: Kluwer Academic Publishers.
Merriam, G. 1984 Connectivity: a fundamental ecological characteristic of landscape pattern. Pp. 5–17 in J. Brandt and P. Agger, eds., Proceedings of the First International Seminar on Methodology in Landscape Ecological Research and Planning. Denmark: Roskilde University Center.
Sicco Smit, G. 1988 A practical application of radar imagery for tropical rain forest vegetation mapping. Pp. 249–264 in A.W. Küchler and I.S. Zonneveld, eds., Vegetation Mapping, Handbook of Vegetation Science. Dordrecht, The Netherlands: Kluwer Academic Publishers.
Sombat, M., and D. van der Zee 1987 The monotoring of Bangkok's rural urban fringe. Ekologiea 6(1):63–76.
van der Zee, D., and J.A. Cox 1988 Monitoring in Moneragala district, Sri Lanka. ITC Journal 3:260–271.
Zonneveld, I.S. 1988 Interpretation of remote sensing images. Pp. 65–68 in A.W. Küchler and I.S. Zonneveld, eds., Vegetation Mapping, Handbook of Vegetation Science. Dordrecht, The Netherlands: Kluwer Academic Publishers.
1989 Scope and concepts of landscape ecology as an emerging science. Pp. 1–20 in I.S. Zonneveld and R.T.T. Forman, eds., Changing Landscape: An Ecological Perspective. New York: Springer-Verlag.