Mapping Soil Classof Depth to Concretions and Mottles Using a Soil-Landscape Model

傳統的土壤調查圖，其繪圖方式乃根據截線路徑的地面採樣點，藉空照圖、地形圖與地質圖的輔助，由經驗豐富的土壤調查員繪製出土壤圖。說明。本研究應用多變值區別分析建立土壤地景模式，整但是傳統的土壤繪於土壤與地景間的關係卻很少作明確的系統之空間分析與土壤性質關係，用以繪製土壤類別圖、土壤類別機率圖與最大機率圖。研究資料取自桃園縣南崁地區土壤性質中鐵、錳結核與灰斑出現深度之調查資料。結果顯示應用土壤地景模式在礫石或砂石母質下生成之土壤，得到的最大機率分佈優於在紅壤母質下生成之土壤；對於鐵，錳結核出現深度的推估也優於對灰斑出現深度的推估，可以得到較高最大機率分佈。土壤地景模式的繪圖技術，除了可繪製結核或灰斑在土壤剖面出現深度的土壤類別圖外，也可以得到相對應的土壤類別機率圖與最大機率分佈圖，並可用以評估土壤類別圖之調查品質，它可清楚說明繪製土壤圖時的不確定性，並且可以快速地更新模式與重繪土壤圖，網格地理資料的儲存也可與其他自然資源應用者共享資料。 Conventional soil maps are made by trained soil surveyors based on observations gathered along transects determined from aerial photos and topographic and geologic maps. The relationships between the soil properties and landscapes are seldom studied in conventional soil surveys. This study used a soil-landseape model developed through multivariate discriminate analysis to integrate the spatial analysis of a geographic information system (GIS) and soil properties, and to establish a new technique for producing soil class maps, soil class probability maps and maximum probability maps. Databases for the depth to Fe concretions, Mn concretions, and grey mottles of the soils as well as the soil drainage class were used in this study. The site was located in the Nankang district, Taoyuan, northern Taiwan. Results showed that the frequency of high maximum probability was higher for soils derived from gravel or sand parent material than for soil derived from red earth material. The frequency of high maximum probability for predicting the depth of Fe and Mn concretions was also higher than that for gray mottles. Application of the soil-landscape model delineates not only the soil class map but also the soil class probability map and maximum probability map. The model could also be used to evaluate the quality of soil maps, to explain the uncertain areas in such maps, and to easily and quickly update them. The raster type of geographic information data created in this study can also be shared with researchers involved in other GIS applications in the natural sciences