| Summary: | 博士 === 國立臺灣大學 === 農業化學研究所 === 90 === The soil map was produced based on the soil occurrence model in traditional soil survey, but the establishment of soil occurrence model requires a big budget, many soil surveyors, and time. The techniques and their applications of geographic information system (GIS), digital elevation model (DEM), and terrain analysis have been completely developed to link the spatial soil database and description database, to investigate the relationship of soil and terrain attributes, and to produce more efficient and reliable soil maps in last few years. The soil-landscape models of soil characteristics and soil maps of major soil groups created by this model are less studied in Taiwan in last decade. The study area is located in Yangmingshan Volcanic National Park in Taipei, including 1600 ha distributed in Tatung Mountain area and 400 ha in Chishing Mountain area. The objectives of this research are (i) to use the terrain analysis and GIS to establish the distribution of major soil groups in study area, and (ii) to establish the regression model with soil and terrain attributes to predict the spatial distribution of soil characteristics.
The results indicate that the collected soils are classified into two Soil Order, two Soil Suborder, four Soil Great Group, and or Soil Subgroup. Furthermore, we established 11 soil series in study area. A soil map of study area was produced based on soil genesis model, terrain analysis and ARC/ INFO GIS (scale 1/25,000). The results of overlay analysis indicate that the similarities of soil series or soil great groups distributing in 90% of total survey areas ranged from 49 to 97% (73% in average), and that of 10% of total survey areas ranged from 27 to 80% (47% in average), excluded miscellaneous area. In conclusions, the terrain analysis derived soil map exhibited reasonable similarities (60% in average) compared with the traditional soil map produced in this study. Soil mapping by selection of transects, terrain analysis and GIS can not entirely replace the traditional soil survey mapping, but it is a useful, time efficient, and cost-effective tool to produce a soil map in large study area.
In large area, it is difficult to establish the soil-landscape model and to predict the spatial distribution of soil properties only based on limited soil sampling numbers. The soil-landscape model and the regression model which we established were strongly influenced by the selection of transects and the representative of sampling location. The result also indicated that the northeastern monsoon strongly affected the genesis and distribution of volcanic soils in Yangmingshan National Park. It needs further study and investigation in the future.
|
|---|
