Geography

The peasant population of feudal Japan has been regarded as stagnant based on low fertility and offsetting mortality, an exception to the classic model of demographic transition. However, the rate of population change among Tokugawa peasants varied temporally and regionally. The Tohoku region in northern Japan experienced a total population decline in the eighteenth century while southwestern regions saw a continual growth supported by high fertility. The population decline in eastern Tohoku was not simply the result of the unfavorable environment or scarcity of land but was also influenced by the unique family structure that essentially determined the distribution of village resources. Facing population pressure, the dozoku groups required a systematic control of reproduction for the survival of their own group. This was accomplished through the regulation of marriage and creation of branch families, urging temporary out-migration, and possibly infanticide.

This thesis examines the quantification of tropical deforestation, the use of remote sensing techniques for its scientific measurement, and the many controversies surrounding the problem. Aerial photographs and Landsat-based planimetric maps were used to determine the conversion of montane rain forest in a 1,000 km$\sp2$ sector of Peru's Huallaga River Valley. Between 1963 and 1976, 244 km$\sp2$ of forest (approximately a quarter of the study area) were converted to agricultural and other land uses, an apparent deforestation rate of 19 km$\sp2$/yr or approximately 1,872 ha/yr. The method entailed the cutting and weighing of strips of Mylar overlays. Despite the photogrammetric limitations, the results demonstrate an economical and practical technique that is readily applicable to developing countries. The potential of other remote sensing systems and the application of change detection techniques such as digital image subtraction to monitor deforestation is detailed.

Patterns of settlement are examined using a number of fractal and scaling techniques. Existing methods of spatial analysis are reviewed and evaluated, showing a need for additional methods for the analysis of spatial and geographical data. Fractal methods are introduced as candidates for such uses. They are applied in a variety of ways to analyze patterns of population and settlement: in an exploratory fashion, to test hypotheses, to determine correlations, to validate settlement models, and to explore the dynamics of population. Scaling properties of population and settlement patterns averaged over different regions are found to be non-uniform, allowing comparisons to be made between them. Correlations are discovered between the scaling behavior of settlement patterns and physical factors such as proximity to coastlines and the ruggedness of the landscape. Popular stochastic and central place models are shown to be inadequate for explaining the scaling behavior of real settlement patterns. Finally patterns of population change are examined, compared and described in terms of their scaling properties. This paper demonstrates how fractal and scaling methods can be usefully applied in the social and spatial sciences. Such an exploratory analysis of the scaling behavior of patterns of human population and settlement constitutes a necessary first step toward a detailed understanding of these phenomena.