Simulations in geography

Abstract

This paper first discusses the responsibilities of geography as a social existence and its contributions to society. Geography must stimulate intellectual curiosity, especially that of scientists in related fields, and geographers must report on their specialty using common scientific concepts. A map is a model of the actual world. Therefore all doings on maps should be called simulations in the broad sense of the term. We cannot suppose better commercial opportunity findings, location decisions and military operations without using maps. These simulations or processes of geographic thinking have become so common that is not realized that they are simulations. Geographic simulations have been renewed by digitalization of maps or GIS. A geographic simulation of vegetation zo ning in Japanese Islands was initiated based on my previous studies. Using vegetation and climate grid data with 1-km resolution, I sought threshold values of thermal conditions that separated forest types of natural vegetation remaining such as evergreen broad-leaved, deciduous broad-leaved, evergreen needle-leaved, and alpine dwarf. Based on these values, I estimated potential vegetation for sites where native vegetation had been destroyed by human activities, and also those under climate conditions 7°C lower 2°C higher than at present. The spatial correlation between the present geographic distribution of vegetation and its natural limitations is a key to the past and also to the future. Two additional simulations in narrow or common sense were based on previous studies. One was on landform development of a small drainage basin. The fundamental equation (model) for the slope was a partial differential equation like as the heat conduction equation, and that of a river profile was the same but used a nonconstant diffusion coefficient, which is an exponential function of distance. Thus the drainage basin was divided into the two domains of slope and fluvial process. This scheme enabled the two-dimensional model to accept a set of rock control and climatic influences as the diffusion coefficient, and sea level change as a boundary condition. The simulation was run on a regular hexagonal DEM to determine future landforms on the assumption that climate and sea level change would continue for the future 100 ky in the same manner as in the past 100 ky. The other simulation was of a probabilistic model of a drainage network. The network system is fundamentally regarded as a binary tree. The simulation models were composed of recursive functions that generated a network by adding a new branch after a random number (Monte Carlo method) and calculated properties of the network such as drainage area, Horton-Strahler's orders, bifurcation ratio, etc. In this case also, the regular hexagonal DEM provides greater convenience in the simplicity of model algorithms. Explaining geography using the word "simulation" is the best strategy for acquainting the public with the field, because sciences are evaluated based on their prognostic ability.