U.S. flag

An official website of the United States government

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

National Research Council (US) Panel on New Research on Population and the Environment; Entwisle B, Stern PC, editors. Population, Land Use, and Environment: Research Directions. Washington (DC): National Academies Press (US); 2005.

Cover of Population, Land Use, and Environment

Population, Land Use, and Environment: Research Directions.

Show details

6Population Change and Landscape Dynamics: The Nang Rong, Thailand, Studies

Stephen J. Walsh, Ronald R. Rindfuss, Pramote Prasartkul, Barbara Entwisle, and Aphichat Chamratrithirong

INTRODUCTION

During the past 40 years there have been two major popular and scientific environmental initiatives. The first was in the 1960s and early 1970s, perhaps triggered by the 1962 publication of Rachel Carson's Silent Spring. The second focused on global climate change and is perhaps best indexed by the formation of the International Geosphere-Biosphere Program in 1986.

During the first environmental research initiative, population researchers tended to stay on the sidelines. There were several reasons. First, the research questions were not well framed; they were more polemical than scientific. Second, appropriate, high-quality data to address the environmental concerns did not exist. Third, many of the necessary methods had not yet been developed. The research community using remote sensing and geographic information systems (GIS) techniques also was not actively engaged in this first round of environmental concern and research. Again, the reasons were varied. First, satellite (Landsat) data became available only in the early 1970s. Initial research with Landsat data focused on broad mapping issues that only suggested pattern-process relationships, including the human dimensions of land use and land cover and subsequent changes in landscape patterns. Such early studies were without a clear and rigorous link between people and the environment; the exploration of possible feedbacks between human behavior and environmental patterns, processes, and dynamics was yet to come. Also, in these early years of mapping land use and land cover, the remote sensing community was dominated by natural and spatial scientists; therefore, the applications of remote sensing technology tended to emphasize resource assessments and far less on social sciences and its importance in understanding the role of people as agents of environmental change and land use and land cover dynamics.

Population researchers and the remote sensing and GIS research community became active during the second initiative. The Nang Rong projects, which began in the early 1980s, began to focus on land use and land cover change partially in response to questions about global warming, environmental degradation, and human behavior. The links between global warming and land cover change, especially deforestation and reforestation, were in the process of being established (e.g., Meyer and Turner, 1992; Kasperson et al., 1995; Houghton et al., 1999; Lambin et al., 1999). We were primarily a team of sociologist-demographers, adding environment, geography, and GIS science expertise in the early 1990s. The issue of land use and land cover change fit our preexisting theoretical concerns. Global change issues provided additional impetus.

In this chapter, we describe the portion of our ongoing work in Nang Rong, Thailand, relevant to the human dimensions of global environmental change, with an emphasis on mapping and modeling patterns and dynamics of land use and land cover by linking people, place, and environment in fundamental ways to address research questions that extend across the social, natural, and spatial sciences and that require integration of data, methods, and perspectives.

QUESTIONS IN CONTEXT

What are the reciprocal relations between population change and landscape dynamics? How do these relations operate at different social, spatial, environmental, and temporal scales? What are the scale, pattern, and process relationships that extend across social, biophysical, and geographical domains? These are the large questions that motivate our research. We have not tended to begin with substantive environmental questions and then create links to people and place. Rather, we have posed basic questions that seek to understand how social change and environmental change are linked. Adapting the research questions to our research setting, the scales at which we work, and the information and tools that are either available or possible to devise, we have focused on migration and household formation as engines of population change, as well as on deforestation, the expansion of rice production, the resource endowments of sites, geographic connections between places, and the introduction of upland crops as fundamental aspects of land use and land cover change. We have studied these issues at seasonal, annual, and decadal scales and at fine to coarse spatial scales over the past half-century, and we have considered the relationships between people, place, and the environment as well as the feedbacks among human behaviors, geographical settings, and ecological processes.

Nang Rong district, Thailand, is our research setting (Figure 6-1). The district occupies approximately 1,300 square kilometers (km2) in the northeast region of Thailand. The district and the surrounding region are noted for the undulating landscape. In the lower elevations there is paddy land, and in the higher elevations there is open-canopy forest and uplands for dry field crops (Fukui, 1993). Elevation matters. Relatively small variations in elevation result in major differences in crop suitabilities in this setting. The region is classified as a tropical dry forest characterized by dry dipterocarp forest and woodlands. The environmental setting is one of marginality: low soil fertility, insufficient and unpredictable precipitation, insufficient drainage, and, generally speaking, a limited natural resource base.

FIGURE 6-1. Study area location: Nang Rong district, northeast Thailand.

FIGURE 6-1

Study area location: Nang Rong district, northeast Thailand.

Until the middle of the twentieth century, Nang Rong was a frontier area,1 more similar in this respect to recently settled areas in Latin America than to a typical Southeast Asian setting. Nang Rong was populated through migration combined with high rates of natural increase in the 1950s and 1960s. Even after the frontier closed in the early 1970s, the population continued to grow until the early 1990s. Deforestation has been extensive. Initially, forest was converted to paddy rice in the lowlands. In the late 1960s and through the 1970s, in part because of changed import regulations in Europe, cassava cultivation became profitable, and forest was converted to agriculture in the uplands. At approximately the same time, a paved road linking the district to Korat (a regional city) and ultimately Bangkok was constructed (for military reasons related to the Vietnam War and to communist insurgencies from nearby Cambodia). The interaction of population and environment through forces both within and exogenous to the region has created a dynamic landscape mosaic. Plate 4 shows the district in 1972-1973 and 1997. Changes in the composition and spatial patterns of land use and land cover, principally caused by deforestation of the uplands and the cultivation of cassava or sugar cane or both, are most obvious, but more subtle changes in the lowlands, caused by pond development, loss of isolated trees and small clusters of trees, and the expansion of Nang Rong town, the central market and administrative town (shown in light blue), are also seen in the image comparison.

PLATE 4. Image comparison of land use and land cover in Nang Rong district (bounded by dark irregular line) and a surrounding 10 km buffer area, 1972-1973 and 1997.

PLATE 4

Image comparison of land use and land cover in Nang Rong district (bounded by dark irregular line) and a surrounding 10 km buffer area, 1972-1973 and 1997.

Most villagers today are farmers, growing rice in the extensive lowlands and such crops as cassava, sugar cane, kenaf, and corn in the uplands. The timing and amount of the yearly monsoon is of particular significance. Most of the year's precipitation in Nang Rong occurs as unevenly distributed torrential rains from June to November. Rain almost never falls in December (Konrad, 2000). Rice must be harvested soon after the rain stops and before the fields dry out (Edmeades, 2000). This is a period of peak labor demand in Nang Rong. Migrants tend to leave the district after the rice harvest. It appears that many migrants are seasonal, returning in the summer to help with rice planting. Seasonal pulses in rainfall thus inspire a seasonal pattern in migration. There is substantial interannual variability as well. Annual precipitation totals have decreased over the past 30 years (Konrad, 2000) and there is evidence that this is related to deforestation (Kanae, Oki, and Musiake, 2001). A significant portion of the decrease occurred during the wettest month of the year, October. Monthly precipitation amounts increased slightly during April and May, the period immediately prior to the rice-growing period. June, a critical month for planting rice, experienced a slight decrease in precipitation amounts (although there is a slight increase in the number of days observed for all precipitation thresholds, particularly the number of days having precipitation totals of 1 and 2 inches). There is considerable variability in the annual precipitation totals. Three of the four wettest years occurred prior to 1984 (with 1983 the wettest since records began to be kept in 1965), whereas all four of the driest years occurred between 1989 and 1994 (with 1989 the driest on record). We expect that when the monsoon is late, light, and sporadic, return migration will be less and out-migration greater.

Migration patterns also depend on opportunities elsewhere. Through the 1980s and 1990s, the Thai economy grew at a remarkable rate. This growth was concentrated primarily in Bangkok and the eastern seaboard, as well as in the manufacturing and service sectors. This trend, in combination with the closing of the frontier in the northeast, encouraged young adults from that region to migrate to cities, either temporarily or permanently. There was yearly variability in migration trends (Jampaklay, 2003). In all likelihood, this reflects variability in the demand for labor in the urban areas as well as variability in the monsoon in rural areas. Of particular interest are the potential consequences of the economic crisis of 1997, when the Thai baht was devalued.

Social as well as biophysical phenomena are related to pulses observed on the landscape. For instance, circular migration patterns of young adults working in Bangkok, the eastern seaboard, and other urban places are related to the efficacy of monsoonal rains for the cultivation of lowland paddy rice in Nang Rong district and the selected mode of rice cultivation, broadcast or transplant, which have very different labor requirements. Feedbacks between the monsoon, characterized by floods, droughts, delayed rains, or normal to near-normal rainfall conditions, have implications for people and the environment. Also, upland deforestation and the cultivation of cassava or sugar cane in the district are related to population-environment interactions involving globalization and the emergence of a market economy, travel distances from the nuclear village, population density and land competition from nearby villages, and site conditions, including resource endowments. Finally, the economic status of villages and the nature of household assets are also part of the calculus of land use and land cover dynamics, household decision-making processes about the land, and defining the geographic “reach” on the land through use or ownership patterns.

THEORETICAL APPROACHES AND PERSPECTIVES

Questions about population, land use, and environment are inherently multidisciplinary. The theoretical approaches taken in any application tend to reflect the disciplinary background and training of the research team. In our case, we have looked to the disciplines of geography and sociology, as well as the emerging area of complexity theory, for guidance in our research. From geography, we have drawn on three perspectives: landscape ecology (e.g., Forman and Godron, 1986), human ecology (e.g., Johnston, Taylor, and Watts, 1995), and political ecology (e.g., Blaikie and Brookfield, 1987). From sociology and social demography, we have also drawn on three perspectives: the multiphasic theory of change and response (Davis, 1963), models of household decision making (Stark, 1991), and the life course perspective (Elder, 1974Elder, 1998). Complexity theory is not tied specifically to either discipline, but rather is emerging in both (e.g., Malanson, 1999 e.g., Malanson, 2002; Urry, 2001). We use the complexity perspective as the basis for linking people, place, and the environment to study nonlinear relationships, feedback mechanisms, and critical thresholds related to self-organization and complex adaptive systems. Each approach has informed some part of our research and has provided valuable insight. In this chapter, we reflect on two: the life course perspective and the complexity perspective. We choose the life course perspective because, although central in our work, it is relatively new in research on land use and land cover change. The complexity perspective is also relatively new, both to our research and to the field more generally.

Life Course Perspective

Individuals and their personal histories are at the core of the life course perspective. The life course refers to a sequence of socially defined, age-graded events and roles that individuals enact over time (Elder, 1998:941). Core elements are role states, transitions, and trajectories. Each transition combines a role exit and entry (e.g., from nonparent to parent). Transitions vary in their structuredness or degree of external regulation, duration, timing, predictability, and novelty (Elder, 1998:957; also see Rindfuss, 1991). Trajectories are composed of role states and transitions; trajectories, in turn, provide some of the context for particular transitions (e.g., whether early or late). Central to the life course framework is the notion that individual lives are independent, connected through social arrangements including households and families. Also central is the notion that lives are lived in a historical context, which shapes both the opportunities and constraints faced by individuals as they move through their life course. In the life course perspective, concepts of time range from biological age to historical context. Potentially relevant social contexts range from the micro to the macro.

Although myriad role transitions are of interest in the life course perspective, the transition to adulthood is particularly important to an understanding of population and land use. Young people on the threshold of adulthood make decisions about migration, marriage, household formation, and childbearing. In fact, these all-important behaviors from the standpoint of demographic process are relegated to a relatively narrow segment of life, from the teens to the 30s, depending on context (Rindfuss, 1991). In Nang Rong, young people generally finish their schooling at age 12; few go beyond a primary education.2 Then, they work: cultivating rice and other crops in their home village; as temporary laborers in agriculture and construction in the district and other nearby places; and as factory, service, and construction workers in major urban destinations, such as Bangkok. Often young people rotate among several jobs and locations, on a seasonal or longer term basis. Work and migration experiences during the teen years and early 20s affect whether, when, and where young people marry and form new households. In 1990, the singulate mean age at marriage in Thailand was 23.0 for women, 25.7 for men (Limanonda, 1992; Phananinarai, 1997). Fertility rates are highest for women in their 20s (Hirschman et al., 1994), as they are in most places. It is clear that the behavior of young people is central to relationships between population dynamics and land use. Yet this simple observation is oddly missing from most accounts of these relationships.

A related observation is that environmental impact will depend importantly on the size of this cohort. As Hunter points out, because of the history of mortality decline and lagged fertility decline in most parts of the world, we are now witnessing the largest-ever generation of young people on every continent but Europe (Hunter, 2000:28). The ultimate impact of these large numbers is yet to be understood. To date, the Nang Rong data have featured the 1984-1994 decade, although the addition of another wave of data collected in 2000 has expanded this window of intensive research. Who were the young people most likely to finish school, migrate (out and perhaps back), marry, form households, and have children over the 1984-1994 decade? Roughly speaking, these were people ages 8-25 in 1984, born between 1960 and 1976, and demographically the most significant cohort in recent Thai history. The large size of this cohort potentially changes the context of decision making for all of its members. Cohort size is a key element of historical context that needs to be recognized when drawing general conclusions, not only from the Nang Rong study but most studies currently under way in different parts of the world.

Complexity and Hierarchy Perspectives

The goal of complexity is to understand how simple, fundamental processes can be combined to produce complex holistic systems (Gell-Mann, 1994). Such systems contain more possibilities than can be actualized, and their descriptions are not constrained by an a priori definition (Luhman, 1985). Nonequilibrium systems with feedbacks can lead to nonlinearity and may evolve into systems that exhibit criticality, or phase transitions, a condition in a system in which any outcome is possible, response to perturbation is of any size, and correlations extend across scales (Malanson, 1999). In the Thailand setting, multiple stakeholders interact through endogenous and exogenous processes to create a dynamic land cover and land use system that is space and time dependent, in which feedbacks among human activities, land cover and use change, and ecological dynamics produce nonlinearity. Critical points in the spatial structure of land use patterns and feedbacks have produced a system with potential alternative states and dynamics characterized by phase changes. For example, we have found that as patterns of land use and land cover become more fragmented, young adults out-migrate to engage in off-farm employment, suggesting that the landscape has exceeded a real or perceived threshold of land availability or contiguity for use or ownership, or for subsequent household formation (Rindfuss, Walsh, and Entwisle, 1996). But as a consequence of off-farm employment, households may accumulate additional assets through remittances, thereby affecting household decisions about the use or ownership of the land.

Hierarchy theory, developed in general systems theory and incorporated into ecology to describe the structure of ecological systems through their spatial and temporal organization, is integrated in the context of complexity theory by considering scale-dependent mechanisms and processes operating at fine and coarse grains and extents that give meaning to the characteristic space and time scales being studied (Allen and Starr, 1982; Ahl and Allen, 1996). Fundamental questions being examined revolve around (1) rates, patterns, and mechanisms of deforestation, agricultural extensification, village settlement patterns, and feedbacks among land use patterns and social, biophysical, and geographical processes; (2) spatial and temporal patterns of road development, migration and household formation, land tenure, monsoonal variability, agricultural intensification, cooperative water use, shifts in world markets, and electrification and spread of consumerism as critical thresholds and feedback mechanisms that alter the trajectories of land use patterns; and (3) uncertainties among land use patterns, processes due to system dynamics, and spatial simulations developed through cellular automata and agent-based models. We routinely explore changes in the geographic extent (e.g., size of village territories, regional subsets of Nang Rong district, and the entire district) and grain size (e.g., spatial or temporal resolutions of analysis) for considering population-environment interactions. For example, Walsh et al. (1999, 2001) examined the variations in landscape greenness by linking variables describing people and the environment measured at a range of spatial scales, and Walsh et al. (2003) reported on a similar study in which intraannual, interannual, and decadal scales were examined through variables describing demographic characteristics of villages, terrain settings, and local resource endowments.

MULTIPLE UNITS, SCALES, THEMES, AND PERSPECTIVES

The Nang Rong project database combines and integrates information on many kinds of units, from individuals to households to villages and from pixels to plots, lines, patches, watersheds, and landscapes. It covers a range of spatial, social, and temporal scales of people, place, and environment. It is multithematic, representing social, biophysical, and geographical domains. The design and collection of the relevant data and their integration into a wide-ranging and flexible, spatially explicit GIS database is one of the accomplishments of the Nang Rong projects. Figure 6-2 summarizes the central elements of this database, which are briefly described below.

FIGURE 6-2. Longitudinal social survey data and remote sensing imagery collected for the Nang Rong projects.

FIGURE 6-2

Longitudinal social survey data and remote sensing imagery collected for the Nang Rong projects.

Social Surveys

The social surveys were the starting point for research in the Nang Rong setting. They consist of three waves of data collection: 1984, 1994, and 2000. Two surveys were conducted in 1984: a community survey in 51 study villages and a complete household census conducted in the study villages, with the census obtaining information on all members of all households. The census design in the context of a highly focused and largely local study laid the groundwork for new insights about demographic behavior (Entwisle et al., 1996Entwisle et al., 1998), subsequent innovations in data collection (Rindfuss et al., 2003Rindfuss et al., 2004), and, importantly, the integration of demographic, biophysical, and land use and land cover data at social, spatial, and temporal scales that played to strength on the social demographic as well as ecological sides (Entwisle et al., 1998).

A second round of surveys were fielded a decade after the baseline, building on and extending the original design and focus. The 1994-1995 data were collected through a community survey administered in all villages in Nang Rong, including but not limited to the original 51; a household survey; a complete census of all households in each of the 51 villages; and a migrant follow-up, which collected data from out-migrants from 22 of the original 51 villages who had gone to one of four urban destinations (metropolitan Bangkok; the eastern seaboard, a focus of rapid growth and development; Korat, a regional city; and Buriram, the provincial city).

The 2000-2001 data collection included a community survey in all villages in Nang Rong; a household survey; a complete census in the 51 study villages; and the collection of locational data for dwelling units and agricultural plots, linked to the household survey, as well as a migrant follow-up that tracked migrants from 22 villages to the four urban destinations and to rural villages in Nang Rong district. Because of administrative subdivisions of villages (villages are generally administratively divided when the total number of households is greater than 100), the original 51 villages expanded to 76 in 1994 and 97 by 2000. Also, because of village subdivisions, the total number of administrative villages in the district expanded from 310 in 1994 to 346 in 2000.

Remote Sensing

An aircraft and satellite image time series has been assembled that extends from 1954 to the present. Panchromatic aerial photography at scales ranging from 1:6,000 to 1:50,000 have been acquired for 1954, 1968, 1969, 1974, 1976, 1982, 1983-1984, 1985, and 1994. Digital air photo mosaics have been derived for selected periods and reformatted into seamless image mosaics that cover the entire district. The digital air photo mosaics are single high spatial resolution images that are easily combined with other digital data contained in the GIS database and integrated as part of overlay analyses and data visualizations. In addition to extending our remote sensing time series to deeper historical periods, the digital mosaics, particularly those that correspond to field or social and demographic surveys, are used to validate classifications of land cover. We are also using the high resolution digital aircraft data to calibrate and validate statistical and spatially explicit models.

Our primary remote sensing platform and sensor system is the Landsat Thematic Mapper (TM). This system “views” the landscape at a 30 m cell resolution for its optical channels, and offers relatively large geographic coverage for each collected image or scene (i.e., 185 by 185 km). Both the grain and extent of Landsat are useful for assessing human imprints on the land indicative, for example, of population settlement patterns, land clearing for the cultivation of crops, and a host of other land transformations that have important population-environment signatures and cause and consequence implications for examining land use and land cover dynamics.

Approximately 35 images have been acquired from the Landsat TM for the period 1973-2003. SPOT (Le Système pour l'Observation de la Terre (Earth Observation System) panchromatic (10 m spatial resolution) and multispectral (20 m spatial resolution) data have been acquired for selected high interest dates approximated to social survey periods. Also, Ikonos data3 are being used to assess settlement patterns of Nang Rong district migrants who have sought off-farm employment in Bangkok. Using migrant data, destinations of migrants from 22 survey villages (1994 and 2000 surveys) are examined in the context of urban morphology and evolution, including core-periphery concepts, migration streams, and patterns of “new” and “old” migrants and issues related to social networks. The Ikonos data provide both multispectral and panchromatic images with very high spatial resolution: the multispectral imagery has a 4 m spatial resolution, whereas the panchromatic imagery has a 1 m spatial resolution.

Using the deep Landsat TM time series, a number of image change detections are under way. Similar to a panel data set familiar to social scientists, we are tracking the “pixel histories” or trajectories of co-registered, classified images to explore the persistence and dynamics of land use and land cover patterns in both space and time and to associate the social, biophysical, and geographical determinants of patterns of these changes through space and time. Other change detection approaches such as change vector analysis, postclassification change, binary masks, and principal components analysis, are being derived (Walsh et al., 2003).

Geographic Information System

A GIS was also developed that includes a number of base coverages as well as derived coverages for the study area. Most fundamental was the generation of a digital elevation model. Using a 1:50,000 scale 1984 base map from the Thai Ministry of Defense, contour lines and spot elevations were digitized. A 10 m contour interval was used on the 1984 map, and spot elevations were maintained to a 1 m vertical resolution. Using the terrain information and the linear surface drainage patterns (perennial and intermittent rivers and streams and ponds or reservoirs), a digital elevation model was developed along with a number of value-added terrain products (e.g., topographic curvature, topographic convergence or wetness index, and solar radiation potential). Also fundamental to our studies was the development of a road network generated by digitizing road types from the 1984 base maps. Roads were described on the Thai base map (and subsequently digitized) as paved all-weather roads, loose-surface all-weather roads, fair dry-weather roads, and cart paths. Using a derived aerial photo image mosaic for 1954, 1967-1968, and 1994, the road network captured from the 1984 Thai military map has been expanded to examine how changes in geographic accessibility through a dynamic road network may contribute to land use and land cover dynamics throughout the Nang Rong district. The district outline was also captured from the 1984 base map, as well as district villages and regional market towns. In the 1994 and 2000 surveys, villages (and households) were also geographically referenced by using differentially corrected global positioning system coordinates. Discussions with village headmen have yielded important updates to the road network. The hydrographic data layer also is being augmented by satellite views of surface water impoundments to extend the 1984 representation to important antecedent and subsequent periods. Hydrologic flow data at dams and stream gauging stations add to our understanding of rainfall patterns captured at a constellation of climate stations within and surrounding the study district. Climate stations are being used to assess the spatial and temporal patterns of precipitation and droughts and their deviations from long-term normals as an important exogenous shock to population-environment interactions.

MULTIPLE ANALYTIC APPROACHES

Statistical Approaches

Our initial approach to an integrated analysis of population and land use and land cover change was to incorporate measures based on one into a disciplinary framework that already existed for the other. For example, we wondered whether the proximity and relative availability of forest lands would affect whether young people chose to stay or leave the village. Again, in the life course framework, the transition to adulthood is a time when numerous transitions are occurring in both the work and family spheres. In the 1970s and 1980s, land was ambiguously titled in much of Nang Rong, which allowed young people to clear a land parcel, farm it, and claim it as their own. Since for all practical purposes farming was the only viable occupation in Nang Rong, access to land was a crucial issue for young people as they considered their work and family future. We examined the availability of nearby forest cover on the migration patterns of young people. The unit of observation was individuals nested within villages, the outcome of interest was migration between 1984 and 1994, and the statistical set-up a multinomial regression analysis. We incorporated measures of land cover based on satellite data from the 1970s as independent variables at the village level. Specifically, we used measures of forest cover within a set distance from the village (radial buffer) and of land cover fragmentation, and we related them to whether two prospective panels of young persons moved out or stayed in the village. We found that young persons were more likely to stay if they lived in villages close to forest cover and with less land cover fragmentation (Rindfuss et al., 1996). Forest cover is an attractive force with respect to the out-migration of young people.

An example illustrating the incorporation of social variables into a spatially based approach can also be given. We were interested in whether relationships between population variables and measures of land cover were stronger at smaller than larger spatial scales. To answer this question, we had to take measures of population referring to discrete points and distribute them over continuous space (Rindfuss et al., 2002). Although people in Nang Rong live in clusters, their impact on land use extends far beyond their village clusters. Considering relationships between social and biophysical factors explaining plant biomass patterns for a single time period for nine scale steps ranging from 30 to 1,050 square meters, we found that relationships between measures of population (such as number of households) and measures of land cover and land use were stronger at finer than coarser scales, and that relationships between biophysical measures (such as slope) and land use were stronger at coarser than finer scales (Walsh et al., 1999, 2001).

More recent work has focused on land cover change assessed for intraannual (i.e., seasonal), interannual (1994-1995), and decadal (1993-2002) intervals (Walsh et al., 2003). In this work, topographic settings, such as the mean elevation and mean slope angle as well as mean distance to water in a prescribed village territory, were significantly and positively correlated with the percentage of land cover change. This is true for all three time intervals, although the strength of the association varied, suggesting some temporal scale dependency. Distance to Nang Rong town, the central market town in the district, was also significantly correlated with the percentage of land cover change for all three time intervals, but, interestingly, the correlation was positive for seasonal and decadal change but negative for annual change. The price of rice land (a proxy for land quality) was statistically and negatively correlated with the percentage change of land cover for the intra- and interannual periods.

The correlations reported so far all involve an assessment of change versus no change, without regard for the direction of change, if it occurs. Turning now to a specific direction of change, from forest to rice (1993-2002), the data show significant and negative relationships between the growth in the number of households between 1994 and 2000, the density of settlement as reflected in the number of villages located within a 3 km buffer surrounding each village, and the distance to Nang Rong town. An interpretation involves a historical context of prior development to support lowland rice production, thereby creating a relatively stable landscape during the decadal study period, and demographic change, primarily the increase in the number of village households and a corresponding increase in the number of villages formed through administrative splits when the number of households exceeded 100. Off-farm employment is also increasing, particularly in villages that border Nang Rong town, the central market and service center of the district.

Mathematical Models: Cellular Automata and Agent-Based Models

Agent-based models examine the basic characteristics and activities of individual agents as the basic building blocks. Agents differ in important characteristics and their interactions are dynamic, in that the characteristics of the agents change over time as the agents adapt to their environment, learn from experiences through feedbacks, or “die” as they fail to alter behavior relative to new conditions or factors. The dynamics that describe how the system changes are generally nonlinear, sometimes even chaotic, and seldom in any long-term equilibrium. Individual agents may be organized into groups or hierarchies that may influence how the underlying system evolves over time. Complex adaptive systems are self-organized systems that combine local processes to produce holistic systems (Bak, 1998). They are emergent in that macro-level behaviors emerge from the actions of individual agents, as they learn through experiences and change and develop feedbacks with finer scale building blocks. Agent-based models capture the building blocks or processes of an emergent hierarchical system.

We have used agent-based models to simulate the development of villages based on elevation, the distance to nearest water, and distance to the nearest preexisting village. Village agents, representing individual agents, simulate the establishment of villages, and geocategory agents manage the geographic categories of information that influence the locations of simulated villages and respond to requests from the village agents (that we have determined to most influence the location of actual villages). A longitudinal survey was used to define village establishment dates for observed villages, an image animation was used to visualize the pattern of observed village establishment on an annual time step, and a GIS was used to represent topography and road and water networks as village attractors.

Cellular automata are used in our research to simulate land use and land cover dynamics. Cellular automata models are composed of a regular grid of cells each in a finite state that are iteratively updated in discrete time steps. The state of a cell is determined by the states of the neighboring cells in the previous time step. Growth or transition rules allow systems to grow from initial conditions, vary their rates of change, or reverse directions in a recursive sequence of iterations. Rules are determined by (1) observing an existing system using an assembled image time series and transition probabilities; (2) theory-based approaches; (3) variable weights and relationships generated through multilevel models and logistic regressions that integrate a longitudinal survey, GIS coverages, and land use and land cover patterns and change trajectories; (4) constructing functions by building and analyzing data distributions; and (5) an expert system informed through qualitative field techniques involving knowledgeable local informants to associate patterns of land cover change to associated processes. The rules embodied in a cellular automata model can be developed to realistically represent the decision making of multiple actors. These decisions are made in the context of an existing pattern of land use and land cover. Each cell in an area represented by the model begins at a known land cover state. In a true cellular automata model, the transition of that cell to another state is determined by the states of surrounding cells. For highly dynamic land cover types, transition probabilities can be developed that depend on the state of the cell and its surrounding cells on its resource endowments (e.g., slope, soils, vegetation, hydrology), demographic characteristics (e.g., population density and number of households), and its geographic proximity (e.g., roads, markets, water, and other villages). These probabilities are generally derived from the satellite image time series, GIS coverages, and the longitudinal household survey (Messina and Walsh, 2001). Once transition probabilities are determined, the model runs stochastically to increase the number of possible outcomes and to better represent dynamic systems. Model convergence and variable sensitivities of the analysis tracks are compared, and analyses of system dynamics and uncertainty are assessed. The results are calibrated and validated by comparing the composition and spatial pattern of simulated land cover to observed patterns represented in a classified satellite data set for the same annual time periods (Messina and Walsh, 2001). Summary correlations and pattern metrics are used to assess differences between observed and expected land cover on the basis of composition and spatial organization.

Complexity theory concepts of critical thresholds, feedback mechanisms, and hierarchy relationships are infused into the cellular automata models for generating simulations to match observed states or for future periods by allowing the model to iterate within the expected bounds of the defined rules. For example, we have developed transition or growth rules to model rice, forest, and upland field crops (cassava and sugar cane). Another example is explicit geographic and biophysical rules that describe the environmental conditions important in the cultivation of rain-fed, lowland paddy rice. Slope angle, elevation, land forms, distance to perennial rivers and streams as well as large reservoirs and ponds, soil suitability, and topographic relative moisture potential are considered. Data distributions, GIS coverages, and satellite maps of land use and land cover are used to define the spatial pattern of each variable, and change functions are developed to indicate the propensity of cells in paddy rice to remain rice or change to another type of land use and land cover in successive time steps in our model runs.

Figure 6-3 is a generalized flow chart showing how rules could be implemented in our cellular automata model to spatially simulate patterns of land use and land cover, primarily lowland paddy rice and upland field crops using initial conditions, stochastic elements, neighborhood conditions, landscape criteria for characterizing cells, and change directions for each model iteration. The model depicted in Figure 6-3 applies interatively across a grid of cells. Initially, the two land classes are modeled separately. Whether a pixel is in rice or in upland field crops depends on a stochastic element (to allow for spontaneous change), the influence of neighboring pixels, and then landscape criteria, such as elevation, slope, soils, access to water, proximity to villages, etc. Because the two land classes are modeled separately, it is possible for a given pixel to be allocated to both. When this happens, the pixel is assigned a given land use according to preset rules about crop precedence. The entire grid is then updated, and the model moves to the next iteration. Cellular automata models allow us to spatially simulate patterns of land use and land cover, examine likely future land use and land cover scenarios, and examine how social and demographic factors at the household or community levels, as well as exogenous shocks, have altered trajectories of land use and land cover change, resulting in possible shifts in the composition and spatial structure of the landscape, which has implications for human behavior.

FIGURE 6-3. Generalized schematic of cellular automata rules for modeling changes in rice and upland field crops, Nang Rong district, Thailand.

FIGURE 6-3

Generalized schematic of cellular automata rules for modeling changes in rice and upland field crops, Nang Rong district, Thailand.

Qualitative Fieldwork

Throughout, we have used qualitative data collected through semistructured open-ended interviews with focus groups, residents, and key informants. We have used these data to pretest survey items, generate hypotheses, validate observations, and assist with interpretation. We illustrate with an example of each use.

First, following a long tradition in the design of social surveys (Biemer and Lyberg, 2003:363-367), we have conducted small studies using qualitative techniques in the development and pretesting of survey items. When developing questions, we want to ensure that the question wording makes sense to our respondents, and that we and they have a common interpretation of the question. For example, in developing our procedures for the 2000 data collection to link households to the agricultural plots they use, we did extensive qualitative prework to ensure that we used terms that were understandable to the respondents and captured our meaning of agricultural plot (Rindfuss et al., 2003). We selected the Thai term plang to represent “plot.” Not only does this term appear throughout the Thai questionnaire, it appears as well in the English translation. Plang is now part of the team vernacular, used in discussions of data and plans for analysis.

Second, we have used qualitative data to generate hypotheses. For example, recent discussions with people in Nang Rong pointed out the importance of charcoal production in the deforestation pattern that occurred in Nang Rong and in the retention of trees in rice paddy fields for sustained production of charcoal. This is now on our list of topics to investigate with quantitative data.

Third, we see a potential for the use of qualitative methods to ground truth classifications derived from historical remotely sensed images. As noted above, our series of aerial photographs begins in 1954, and our series of satellite images begins in 1972. A challenge we have faced is the validation of classifications of land use and land cover based on these images. The general idea is to find individuals who have lived and farmed in Nang Rong for at least 20 years, preferably longer. Then, taking the respondent to a known place relative to the historical classification and using the insights from social demographic life history data collection methods (e.g., Freedman et al., 1988) to orient a respondent or group of respondents to the time period covered by the remotely sensed image, ask the respondent to recall land cover for particular points in time. This is a variant of a procedure used in Brazil (Moran et al., 1994; Moran and Brondizio, 1998).

Finally, we have used qualitative data collection to help understand and resolve puzzling results that emerge from our models. For example, in recent work using agent-based models to understand the historical pattern of the location and growth of villages, the models were doing poorly in predicting an arc of villages in the southwest corner of Nang Rong. This is an area with little surface water, and the model did not predict the location of villages there. Discussions with older residents of these villages made it very clear that the presence of underground water near the surface led to the founding of villages along this arc. When looking for possible sites to start a village, information about underground springs was obtained by potential settlers from a number of sources. First, the vegetation was distinctively different when water was close to the surface. Second, large mammals would scratch the ground to expose the water for their own drinking. Hunters would discover these places and, in turn, would tell those looking for a village site. There was some timbering occurring in the area, and those looking for an appropriate village site would also learn about subsurface springs from loggers.

INTEGRATING CHALLENGES

Vocabulary

Land change science projects are, by their very nature, interdisciplinary, and ours was no exception, bringing together sociologists, environmental geographers, spatial scientists, and social demographers. Scientific disciplines develop their own language or jargon to facilitate communication among members, but this, in turn, makes it difficult to communicate across disciplines. In our case, this happened with the term “network” or “network analysis.” In sociology, the term implies a social network which links social actors (individuals, households, organizations, and so forth) to one another by links generated through kin relations, friendship, neighborhood proximity, or other generators. For those steeped in GIS science, network analysis refers to travel from one point to another along routes that have assigned travel times, often within a location/allocation context.

Long Distance Collaboration

Since the mid-1980s, researchers at the Institute for Population and Social Research, Mahidol University, Bangkok, and at the Carolina Population Center, University of North Carolina, Chapel Hill, have been collaborating on the ever-evolving Nang Rong studies. Bangkok and Chapel Hill are separated by 12 time zones. This means that there is no overlap in the normal working hours at the two sites, making telephone discussion problematic. In recent years, email connectivity has been excellent, but email is a poor substitute for face-to-face discussions when project research questions, techniques, and analyses are rapidly evolving, and especially when more than one discipline is involved. And the roughly 36-40 hours travel time required for a round trip between Chapel Hill and Bangkok, not to mention the required time zone adjustments, make it daunting to frequently get together for collaborative meetings. These problems can clearly be surmounted by visits and extended stays, but our progress undoubtedly would have been faster if we were all at the same university and the study site was right next door.

Locating Collaborators

As the goals of our projects have become more complex, we have found ourselves needing additional collaborators who have the required expertise that goes beyond our own. Three areas have been important: social network analysis, cellular automata modeling, and agent-based modeling. Relying on a network among our own colleagues, we were able to locate appropriate collaborators at the University of California, Irvine; the University of Iowa; and Michigan State University. These collaborations are working well, but face-to-face meetings have been limited by travel and time costs.

Develop New Data Collection Techniques

At a number of points in our research we found that standard techniques in demography, geography, and sociology did not meet our needs (Rindfuss et al., 2002Rindfuss et al., 2003Rindfuss et al., 2004). Nowhere was this more the case than in linking people to the land units over which they have decision-making power. Early in our research on land cover and use change, we linked at the village level, but this had numerous problems in a district in which people lived in nucleated villages, households had multiple agricultural plots, and land was frequently ambiguously titled (Entwisle et al., 1998; Evans, 1998; Rindfuss et al., 2002). The technique we devised combined interview, focus group, and GIS/remote sensing techniques. The details have been reported elsewhere (Rindfuss et al., 2003), but the basic elements are as follows. For all areas for which cadastral maps existed, they were obtained, digitized, and overlaid on an aerial photograph for use in the field. If cadastral maps were not available for a village, then just the aerial photography was brought into the field. These aerial photos, with or without cadastral lines, were used in focus groups containing village members with specialized knowledge of local land use to mark the parcels used by village households. In the household interviews, households were asked details about the agricultural parcels they used and they were asked to provide information on those who used the neighboring agricultural parcels. Then the spatially explicit focus group information and the relationally specific household information were brought together, by hand, to determine the location of the household plots, using skills similar to those used when working on a jigsaw puzzle (Rindfuss et al., 2003).

Importance of Starting Point

Our project began as an evaluation of a community-based integrated rural development intervention project. Its focus on a single district, a significant proportion of the villages in that district, and all of the people and households in those villages created opportunities for blending diverse data and analytic techniques to address interesting questions about population change and landscape dynamics. However, the specifics of the starting point also put limits on what is possible. For example, the original set of study villages excluded the southwest, the part of the district that has experienced the most dynamic land use change since 1984. Community-level data were collected on villages in the southwest starting in 1994, as well as some historical information included as part of that, but we lack the detail on households, household members, migrants, field plots, and land use available for the study villages. Village histories can be reconstructed (Entwisle et al., no date-a), and village-level analyses are possible, but the scale at which questions can be addressed is limited in this part of the district. As another example, the original design featured a census of households, but in the Nang Rong setting, this does not translate into a spatially continuous set of field plots, even locally, much less complete spatial coverage of the district. Many examples can be given of how our starting points—a social demographic survey and its design features—placed constraints on what was possible on the environmental side. Turning it around, if we had begun our project with the historical aerial photos, we might have proceeded very differently.

CONCLUSION

Our research on Nang Rong is based on a study of a single district in northeast Thailand, covering an area the size of a county in the eastern United States. Other investigators also focus on similarly constrained settings. Interconnections between human actions and the biophysical environment can be most easily observed at the local level, although, even at this level, there can be important spatial and temporal scale dependencies (Walsh et al., 1999, 2003). A focus on local as opposed to regional or global contexts and seasonal, annual, and decadal time periods directs attention to migration and household formation in addition to fertility and natural increase as master population processes with respect to land use and land cover change. A case study approach thus makes sense, given the goals. A case study is an intensive and focused examination of a phenomenon within and with attention to its context, often relying on multiple sources of evidence (Yin, 2003). It is representative only of itself, although it may be possible to draw larger inferences if the case study is embedded with other case studies in some kind of comparative design (Ragin, 1987). In the literature on population dynamics and land use change, case studies are often fairly local.

Our goal is a comprehensive account of social, economic, demographic, and environmental change in Nang Rong district. Our approach is cumulative, with research questions and data intimately connected. One element at a time, we have built a GIS database that is substantively broad, includes multiple levels of observation, and is unusual in its temporal depth. Each new element, whether a data source or an analytic result seems to create the need for other new or expanded elements. For example, village administrative histories were collected in 2000 that made possible a new look at village settlement and expansion over the past 50 years (Entwisle et al., no date-a). New settlement occurred relatively early in the period and was largely complete by 1970, just before our satellite time series begins. This in turn sparked renewed interest in the aerial photos (dating back to 1954) as a way to characterize land cover at the beginning of the period and led to the classification of portions of them. Land cover changes associated with village settlement and expansion now, which include the fragmentation and reconsolidation of the landscape as well as its composition, can be described on a decadal scale (Entwisle et al., no date-a). As the data set has grown in size and complexity, it has also increased in diversity, creating the need for new and more linkages between data of different types at different time and space scales. Our team has been particularly active in the development of linkages between survey-based social units and GIS-based land units (Rindfuss et al., 2002Rindfuss et al., 2003). Increasing knowledge of the case combined with the data we have assembled provides an opportunity for testing a broad range of new ideas and approaches. We have tended not to become engaged in policy issues or policy-relevant interpretations of our findings, but rather to emphasize fundamental research questions and analytical methods in the study of population-environment interactions.

For example, as part of our longitudinal social survey, we have tracked rural migrants to urban places. As a consequence, we are now engaged in a study of Bangkok using geocoded migrant locations and remote sensing to understand urban form and function, migrant destinations, and the scaling of remote sensing systems designed to study local neighborhoods and regional settings. We have expanded the team to include another geographer with urban remote sensing interests. Also growing out of our social surveys, we have linked people to the land that they use in our 2000 round of data collection. Using cadastral maps and expert discussion groups in study villages, we have been able to enhance our prior efforts in setting village territories by now spatially associating households to the land that they use. Borrowing from spatial analysis, we have developed approaches for characterizing the spatial organization of “functional” village territories by generating activity spaces, “radiance” diagrams that show the vector connections between the village centroids and the center of land parcel being used by village household, and triangulated irregular networks for characterizing village territories using facets whose nodes are land parcels in use by village households.

Building on our classifications of nearly 35 Landsat TM images, we are developing land cover trajectories or “pixel histories” so that we can examine space-time patterns of land cover change (Crews-Meyer, 2001Crews-Meyer, 2002). This work has required customized programming in SAS (Statistical Analysis Software) to identify land cover sequences in a panel context for the generation of subsequent models of land cover dynamics that integrate multidimensional and scale-dependent drivers of change. Our models of land cover have also evolved from the statistical to the spatial. Using cellular automata and agent-based models, we are examining the patterns and processes of land cover dynamics by considering nonlinear relationships, emergent behaviors of agents, and complex adaptive systems. Beginning with “pseudo-agents,” we are modeling the expansion of villages over space and time according to rules of behavior and an evolving landscape. Our agent-based models will soon begin to consider households and their decision-making processes that affect land use and cover change patterns and that integrate social, biophysical, and geographic factors and the interactions between agents and their environment. Cellular automata models are being used to simulate the environment (using the satellite time series for model calibration and validation) through patterns of land use and land cover change (set by initial conditions, transition or growth rules, and neighborhood relationships) that the agents use by generating surfaces of locational and biophysical change (Clarke, Hoppen, and Gaydos, 1996; Clarke, Gaydos, and Hoppen, 1997). Exogenous and endogenous factors will be represented in our models through rules and behaviors represented at the building block level (i.e., household). The potential value of integrating GIS, cellular automata, and agent-based technologies for modeling and simulating social and ecological phenomena has been discussed by Westervelt (2002) and Wright (2002) but not yet fully implemented in the study of population-environment interactions.

Findings from our research have contributed to the social, natural, and spatial sciences and have emerged as a consequence of our integrative and cross-cutting perspectives. For example, we have (a) chronicled patterns of land use and land cover change by developing a dynamic model of village settlement that depends on prior settlement patterns and on potential relevant contextual factors (Entwisle et al., no date-a), (b) determined that growth in the population of households is a stronger predictor of upland crop production than growth in the number of persons (Entwisle et al., no date-b), (c) related land use and land cover fragmentation patterns of forests located within village territories to the out-migration patterns of young adults to Bangkok and other urban destinations (Rindfuss et al., 1996), and (d) described the variation in landscape greenness as a function of demographic and biophysical variables that are space and time dependent (Walsh et al., 1999, 2001, 2003).

The challenges are many for interdisciplinary research in population-environment interactions, but the opportunities for insight are substantial. We continue to chart our research by posing questions rooted in theory and practices, but extending across traditional disciplinary borders. We are concerned with questions that link people, place, and environment in a spatially explicit context, in which endogenous and exogenous factors are considered, and in which the multiscale and multithematic determinants of land use and land cover change are examined. We use perspectives fundamental to the social, natural, and spatial sciences and also those that offer new theoretical insights and possible links to evolving analytical methods. Our questions are about human behavior, primarily migration and demographic characteristics at the household level, and the concommittal changes in the composition and spatial structure of the land. Space and time lags in population-environment interactions are examined to further consider the causes and consequences of land use and land cover dynamics. Geospatial data and spatial digital technologies are used to map and model the social, biophysical, and geographical landscapes. As spatial and spectral resolutions of remote sensing systems continue to increase, we will increasingly rely on such systems to add further insights about the land and the people that shape it. But images are only snapshots in time, they do not communicate intentions or memories, and they are only indirect measures of population and can only hint at questions of human behavior. Images are also spatially and temporally biased to certain human imprints on the land. People are mobile, discrete, and their links to the land may be through lands on which they reside or from some more distant location. Remote sensing and the land that it senses is generally static in place but may be dynamic through time, is continuous in pattern, but reflects patterns only at specified scales without providing insights about the motivations of people and how they might translate to actions that affect the land. In our research, it is the integrative power of our theories and perspectives, data and methods, and analyses and interpretations that informs our inquiries and offers insights to complex problems, some newly emerging, while others are cast in terms of traditional paradigms and perspectives, but served by new analytic approaches and syntheses that extend across the social, natural, and spatial sciences.

ACKNOWLEDGMENTS

The research reported here has been supported by a variety of mechanisms, including grants from the National Institute of Child Health and Human Development (RO1-HD33570 and RO1-HD25482). Additional support was provided by the Mellon Foundation (through a grant to the Carolina Population Center), the National Aeronautics and Space Administration (NAG5-6002), the National Science Foundation (SBR 93-10366), the Evaluation Project (USAID Contract #DPE-3060-C-00-1054), the MacArthur Foundation (95-31576A-POP), and a P30 center grant to the Carolina Population Center from the National Institute of Child Health and Human Development, HD05798. We are grateful for this support. We would also like to acknowledge, with gratitude, the help and cooperation of numerous individuals who assisted in the design of the research described here. Numerous staff members and graduate students at the Institute for Population and Social Research, Mahidol University, and the Carolina Population Center, University of North Carolina, participated, as well as colleagues at other universities. And finally, and perhaps most importantly, we would like to acknowledge the cooperation of the people of Nang Rong. Our requests for information have been many, and Nang Rong residents have cooperated completely, and for that we are in their debt.

REFERENCES

  1. Ahl V, Allen TFH. Hierarchy Theory. New York: Columbia University Press; 1996.
  2. Allen TFH, Starr TB. Hierarchy: Perspectives for Ecological Complexity. Chicago, IL: University of Chicago Press; 1982.
  3. Bak P. How Nature Works. New York: Copernicus/Springer-Verlag; 1998.
  4. Biemer PP, Lyberg LE. Introduction to Survey Quality. Hoboken, NJ: John Wiley and Sons; 2003.
  5. Blaikie P, Brookfield H. Land Degradation and Society. London, England: Methuen; 1987.
  6. Carson R. Silent Spring. Boston, MA: Houghton Mifflin; 1962.
  7. Clarke KC, Hoppen S, Gaydos L. Methods and techniques for rigorous calibration of a cellular automaton model of urban growth. Paper presented ath the Third International Conference/Workshop on Integrating GIS and Environmental Modeling, Santa Barbara National Center for Geographic Information and Analysis; Santa Fe, NM. 1996.
  8. Clarke KC, Gaydos L, Hoppen S. A self-modifying cellular automaton model of historical urbanization in the San Francisco Bay area. Environment and Planning B. 1997;23:247–261.
  9. Crews-Meyer KA. Assessing landscape change and population-environment interactions via panel analysis. Geocarto International. 2001;16(4):69–79.
  10. Crews-Meyer KA. Characterizing landscape dynamism using paneled-pattern metrics. Photogrammetric Engineering and Remote Sensing. 2002;68(10):1031–1040.
  11. Davis K. The theory of change and response in modern demographic history. Population Index. 1963;29:345–366. [PubMed: 12335951]
  12. Edmeades J. Rice farming in Nang Rong, Thailand. Chapel Hill: Department of Sociology and Carolina Population Center, University of North Carolina; 2000. Unpublished paper.
  13. Elder GH. Children of the Great Depression: Social Change in Life Experience. Chicago, IL: University of Chicago Press; 1974.
  14. Elder GH. The life course and human development. In: Lerner RM, editor. Handbook of Child Psychology, Volume I: Theoretical Models of Human Development. New York: John Wiley and Sons; 1998. pp. 939–991.
  15. Entwisle B, Rindfuss RR, Guilkey DK, Chamratrithriong A, Curran SR, Sawangdee Y. Community and contraceptive choice in rural Thailand: A case study of Nang Rong. Demography. 1996;33:1–11. [PubMed: 8690134]
  16. Entwisle B, Walsh SJ, Rindfuss RR, Chamratrithirong A. Land use/land-cover and population dynamics, Nang Rong, Thailand. In: Liverman D, Moran EF, Rindfuss RR, Stern PC, editors. National Research Council, People and Pixels: Linking Remote Sensing and Social Science, Committee on the Human Dimensions of Global Change. Washington, DC: National Academy Press; 1998. pp. 121–144.
  17. Entwisle B, Edmeades J, Malanson GP, Podhisita C, Prasartkul P, Rindfuss RR, Walsh SJ. Village Settlement, Deforestation, and the Expansion of Agriculture in a Frontier Region: Nang Rong, Thailand. Carolina Population Center, University of North Carolina; Chapel Hill: no date-a.
  18. Entwisle B, Walsh SJ, Rindfuss RR, VanWey LK. Population and Upland Crop Production in Nang Rong, Thailand. Carolina Population Center, University of North Carolina; Chapel Hill: no date-b.
  19. Evans TP. Ph.D. dissertation. Department of Geography, University of North Carolina; Chapel Hill: 1998. Integration of Community-Level Social and Environmental Data: Spatial Modeling of Community Boundaries in Northeast Thailand.
  20. Forman RTT, Godron M. Landscape Ecology. New York: John Wiley and Sons; 1986.
  21. Freedman D, Thorton A, Camburn D, Alwin D, Young-DeMarco L. The life history calendar: A technique for collecting retrospective data. Sociological Methodology. 1988;18:37–68. [PubMed: 12282712]
  22. Fukui H. Food and Population in a Northeast Thai Village. Hawkes P, translator. Honolulu: University of Hawaii Press; 1993.
  23. Gell-Mann M. The Quark and the Jaguar. New York: Freeman; 1994.
  24. Hirschman C, Tan J, Chamratrithirong A, Guest P. The path to below replacement-level fertility in Thailand. International Family Planning Perspectives. 1994;20:82–87. 107.
  25. Houghton R, Hackler J, Lawrence K. The U.S. carbon budget: Contribution from land-use change. Science. 1999;285:574–578. [PubMed: 10417385]
  26. Hunter LM. Santa Monica, CA: RAND; 2000. The Environmental Implications of Population Dynamics.
  27. Jampaklay A. Ph.D. dissertation. Chapel Hill: Department of Sociology, University of North Carolina; 2003. Migration, Marital Timing, and Mate Selection in the Context of Thailand.
  28. Johnston RJ, Taylor PJ, Watts MJ, editors. Geographies of Global Change: Remapping the World in the Late Twentieth Century. Oxford, England: Blackwell; 1995.
  29. Kanae S, Oki T, Musiake K. Impact of deforestation on regional precipitation over the Indochina Peninsula. Journal of Hydrometeorology. 2001;2(1):51–70.
  30. Kasperson JX, Kasperson RE, Turner BL II, editors. Regions at Risk: Comparison of Threatened Environments. Tokyo, Japan: United Nations University Press; 1995.
  31. Konrad CE. Chapel Hill: University of North Carolina; 2000. Summary Report: Nang Rong Precipitation Climatology. Unpublished manuscript.
  32. Lambin EF, Baulies X, Bockstael N, Fischer G, Krug T, Leemans R, Moran EF, Rindfuss RR, Sato Y, Skole D, Turner BL II, Vogel C. Land-Use and Land-Cover Change Implementation Strategy. Stockholm: IGBP Secretariat, Royal Swedish Academy of Sciences; 1999.
  33. Limanonda B. Nuptiality patterns in Thailand: Their implications for further fertility decline. In: Goldscheider C, editor. Fertility Transitions, Family Structure, and Population Policy. Boulder, CO: Westview Press; 1992. pp. 101–120.
  34. Luhman N. A Sociological Theory of Law. London, England: Routledge and Kegan Paul; 1985.
  35. Malanson GP. Considering complexity. Annals, Association of American Geographers. 1999;89:746–753.
  36. Malanson GP. Extinction debt trajectories and spatial pattern of habitat destruction. Annals of the Association of American Geographers. 2002;92:177–188.
  37. Messina JP, Walsh SJ. 2.5D morphogenesis: Modeling landuse and landcover dynamics in the Ecuadorian Amazon. Plant Ecology. 2001;156:75–88.
  38. Meyer WB, Turner BL II. Human population growth and global land-use/cover change. Annual Review of Ecological Systems. 1992;23:39–61.
  39. Moran EF, Brondizio E. Land use change after deforestation in Amazônia. In: Liverman D, Moran EF, Rindfuss R, Stern P, editors. National Research Council, People and Pixels: Linking Remote Sensing and Social Science, Committee on the Human Dimensions of Global Change. Washington DC: National Academy Press; 1998. pp. 94–120.
  40. Moran EF, Brondizio E, Mausel P, Wu Y. Integrating Amazonian vegetation, land-use, and satellite data. BioScience. 1994;44(5):329–338.
  41. Phananinarai M. Population Changes and Economic Development in Thailand: Their Implications for Women's Status (East-West Center Working Papers 88-11.). Honolulu, HI: East-West Center; 1997.
  42. Ragin CC. The Comparative Method: Moving Beyond Qualitative and Quantitative Strategies. Berkeley: University of California Press; 1987.
  43. Rindfuss RR. The young adult years: Diversity, structural change, and fertility. Demography. 1991;28:493–512. [PubMed: 1769399]
  44. Rindfuss RR, Walsh SJ, Entwisle B. Land Use, Competition, and Migration. Paper presented at the Population Association of America Meeting; New Orleans, LA. 1996.
  45. Rindfuss RR, Entwisle B, Walsh SJ, Prasartkul P, Sawangdee Y, Crawford TW, Reade J. Continuous and discrete: Where they have met in Nang Rong, Thailand. In: Walsh SJ, Crews-Meyer KA, editors. Linking People, Place and Policy: A GIScience Approach. Boston, MA: Kluwer Academic Publishers; 2002. pp. 7–37.
  46. Rindfuss RR, Prasartkul P, Walsh SJ, Entwisle B, Sawangdee Y, Vogler JB. Household-parcel linkages in Nang Rong, Thailand: Challenges of large samples. In: Fox J, Rindfuss RR, Walsh SJ, Mishra V, editors. People and the Environment: Approaches for Linking Household and Community Surveys to Remote Sensing and GIS. Boston, MA: Kluwer Academic Publishers; 2003. pp. 131–172.
  47. Rindfuss RR, Jampaklay A, Entwisle B, Sawangdee Y, Faust K, Prasartkul P. The collection and analysis of social network data in Nang Rong, Thailand. In: Morris M, editor. Network Epidemiology: A Handbook for Survey Design and Data Collection. Oxford, England: Oxford University Press; 2004.
  48. Stark O. The Migration of Labor. Cambridge, MA: Basil Blackwell; 1991.
  49. Urry J. The sociology of space and place. In: Blau JR, editor. The Blackwell Companion to Sociology. Malden, MA: Blackwell; 2001. pp. 5–13.
  50. Walsh SJ, Evans TP, Welsh WF, Entwisle B, Rindfuss RR. Scale-dependent relationships between population and environment in northeastern Thailand. Photogrammetric Engineering and Remote Sensing. 1999;65:97–105.
  51. Walsh SJ, Crawford TW, Welsh WF, Crews-Meyer KA. A multiscale analysis of land use and land cover and NDVI variation in Nang Rong District, Northeast Thailand. Agriculture Ecosystems and Environment. 2001;85(1–3):47–64.
  52. Walsh SJ, Rindfuss RR, Entwisle B, Shao Y, Weiss DJ, McDaniel PM, Pullen RE. Characterizing land use/land cover change in northeast Thailand and analyzing the causes and consequences of landscape dynamics. (CD-ROM.). Proceedings, 30th International Symposium on Remote Sensing of Environment; Honolulu, HI. 2003.
  53. Westervelt JD. Geographic information systems and agent-based modeling. In: Gimblett HR, editor. Integrating Geographic Information Systems and Agent-Based Modeling Techniques for Simulating Social and Ecological Processes. Oxford University Press; New York: 2002. pp. 83–104.
  54. Wright HT. Agent-based modeling of small-scale societies: State of the art and future prospects. In: Kolher TA, Gumerman GJ, editors. Dynamics in Human and Primate Societies: Agent-Based Modeling of Social and Spatial Processes. New York: Oxford University Press; 2002. pp. 373–385.
  55. Yin RK. Case Study Research: Design and Methods. 3rd ed. Thousand Oaks, CA: Sage Publications, Inc.; 2003.

Footnotes

1

“Frontier” is used to represent the past 500 to 700 years. At the time of the Angkor civilization, roughly 802 to 1432, Nang Rong was part of the Angkor empire.

2

It should be noted that this is changing as more secondary schools are opened and as factory employers are requiring workers to have a secondary education.

3

The Ikonos satellite, launched into orbit on September 24, 1999, is a commercial satellite owned and operated by Space Imaging, Inc.

Copyright © 2005, National Academy of Sciences.
Bookshelf ID: NBK22955
Contents

Views

  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (15M)

In this Page

Other titles in this collection

Related information

Recent Activity

ClearTurn OffTurn On

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...