A Land Use Change Study Using
Cellular Automata Jinn-Guey
Lay Key words Spatial modeling, Land Use, Cellular
Automata (CA)Associate Professor, Department of Geography National Taiwan University, PO Box, 23-76, Taipei, Taiwan Tel: (886)-2-23621499 Fax: (886)-2-23622911 Email: jglay@ccms.ntu.edu.tw Abstract Land use change is a major issue of global environment change. The modeling and projecting of land use change is essential to the assessment of consequent environmental impacts. Recent development of cellular automata (CA) provides a powerful tool for the dynamic modeling of land use change. This research adopts the spatial evolution concept embedded in CA and applies it to land use change study in Tansui Watershed. Digital land use data of two separate years were complied and analyzed using GIS software. A computer program was coded to analyze the neighborhood condition of each target cell. Summary of neighborhood conditions may reveal the dynamic process of land use change and thus enhance our understanding on transition rules, the heart of a CA. Introduction Land use activity is a linkage between human and environment. It is like a seesaw carrying human activities and environment on two opposite ends. Human activity of development is a primary driving force for global environmental changes while the changes of environment in return affects land use types. Envisioning the consequent effects of land use changes, IGBP (International Geosphere and Biosphere Programme) and IHDP (International Human Dimension Programme) co-organized a working group to set up research agenda and promote research activity for land use and land cover changes (LUCC). The working group suggested three core subjects for LUCC research, such as: situation assessment, modeling and projecting, and conceptual scaling. The ultimate goal of global change study is to assess the impacts under each possible scenario and suggest preventive actions. The modeling and projecting of land use change is essential for scenario analysis and the assessment of LUCC. Consequently, issues related to data, information, and modeling have attracted many research interests ranging from local authorities to global organizations. The importance of land use change has long been recognized in Taiwan with many research projects undergoing. This paper presents partial result of an undergoing research, supported by National Science Council, for the study of land use change in Taiwan. Cellular Automata and Land Use Change Recent development of GIS technology enhances the analytical power needed for the study of land use and land cover change. Existing land use data are being digitized while much more data are readily created in digital format, i.e., remotely sensed imagery. The development and applications of digital land use database have been successful as documented in many reports (Lay, 2000). With the help of GIS, these data can be easily overlaid and provide information needed for situational assessment. On the other hand, the methodology for predicting and modeling land use change is relatively immature. Land use change is a dynamic spatial process involves complex interactions between many factors at various spatial extents. The complexity of this dynamic process makes the creation of a comprehensive model very challenging. Recent development of cellular automata (CA) brings in a new perspective for land use change. There are increasing interests in using CA for land use study (Batty, 1998; Clarke et al., 1997; White and Engelen, 1997) The concept of cellular Automata is originated from research for a self-replicating machine, a robot type machine equipped with visions and encoded Turing machine that can assemble a copy of itself from component parts (Firebaugh, 1988). Although such machine has not been successfully constructed, an abstract model of this self-replicating system was constructed as two-dimensional cellular automata. Cellular automata may be represented by a set of simple production rules while its outcome may mimic a very complex system (Firebaugh, 1988, p320-323). From the application side, cellular automata are dynamic model that inherently integrates spatial and temporal dimension. CA are composed of four elements as described below (White and Engelen, 2000).
Land Use Change in Tansui Watershed With a population close to 6 millions, Tansui watershed has experienced a dramatic change during the past decades with huge amount of land use and land cover change. Land use data of the research area are available in various formats, such as: historical documents, paper maps, remotely sensed imagery, and digital data. For convenience reason, digital data are chosen as source data for this research. These data are available separately in 84 data files for each year of 1971 and 1977 in Arc/Info vector formats. The data files of each year are merged together into a big one to cover the whole research area. Since vector files tempt to result in erroneous sliver polygons thus false changes, this research converted the vector files into raster format before further process. Original data is composed of 99 categories of land use types. For the purpose of environmental change, they are reclassified into seven categories: settlement, farmland, bare land, forest, river and reservoirs, other water body, and roads. The data of two years are overlaid with each other to identify the land use change. The overlay function is conducted using Arc View's Tabulate Area function. The change is shown in Table 1 and Table 2. As shown in Table 1, the increase of settlement is most significant with a figure near 4% or 1250 hectares. Each figure in Table 2 represents amount of change from 1971 to 1977 for each category. Based on these figures, this research therefore concentrated on investigating the change from farmland to settlement. A computer program is coded for investigating the neighborhood states of the targets cells.
Table 2?Summary of Land Use Change
Analysis The states of neighborhood is summarized based on the pattern of its 8 neighboring cell regardless its orientation or sequences. Taking the example shown in Figure 1, the land use of a central cell changed from farmland to settlement, and its neighborhood are composed of one settlement cell, six farmland cells, and one forest cell. This neighborhood condition is coded as: 1610000, with each digit indicating the number of cells under each land use sequentially. T1 T2 Farmland Farmland Farmland Settlement Farmland Forest = T1 T2
This research adopts concepts of CA to investigate land use change from a spatial evolution perspective. Factor of concern is limited to neighborhood conditions in this research. Although this approach is rather simple, yet the underneath merit lies in its ability in revealing the spatial dynamic of land use evolution. From a quantitative perspective, the prediction of future land use change may include factor of: amount of changes, rate of changes, type of changes, and location of changes. CA alone may be too simplified to predict all characteristics of changes, yet there is great possibility for CA to integrate with other methods. The creation of transition rules is but a fundamental step, yet the most challenging one, in building a comprehensive model of land use change. Findings from this research help reveal the complex process of land use change, as shown in the various changes resulted from a same pattern of neighborhood. Such finding indicates a deterministic approach of CA may not fit with real world situation. The calibration of such deficiency lends itself to many possible topics for further research. Acknowledgement This research is funded by National Science Council, Taiwan, Republic of China (NSC89-2750-P-002-009). The test data are provided by the Council of Agriculture, Taiwan, Republic of China. Reference
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||