The Classification and Interpretation of Korean Soils Using Multivariate Analysis.
Yeon-Kyu Sonn1, Yeon-Tae Jung2, Myung Chul Seo1, Han-Kang Kwak1, Jeong-Gyu Kim3, Ki-Cheol Eom1, and Soo-Kil Lim4. (1) National Institute of Agricultural Science and Technology, Suin-ro 150, Seodun-dong, Suwon, South Korea, (2) GyeongNam Agricultural Research and Extension Services, 1085-1 Chojeon-dong, Jinju, South Korea, (3) Div. Env. Sci & Ecol. Eng. Korea University, Anam-Dong, Seungbuk-Gu, Seoul, 136-701, South Korea, (4) Korea University, Anam-Dong, Seungbuk-Gu, Seoul, South Korea
Korean detailed soil surveys have been already finished until 1995, and highly detailed soil surveys for upland have been done from 1995 through 1999. Nowadays, the digitalized soil survey results are open to the public on the Internet. Subsequently, soil survey results that were obtained about 30 years ago were replaced with the new ones since 2000. Owing to our efforts, we are at the position of a leading country in the world in the aspect of soil survey and classification, but in spite, we are not yet satisfactory in the aspects of classification and a new interpretation for practical use. Therefore, Korean soil survey data were analysed in order to give new interpretation using multivariate analysis. As these data have no dependent variables and non-quantitative characteristics, so multivariate analysis methods like Multi-Dimensional Scaling (MDS) as possible tool was selected in the analysis. Firstly, collected data on soil survey were standardized then were analysed in order to understand and interpret their properties. And, Korean soils were classified into 19 catenas based on the characteristics of parent rocks and were re-analyzed, and the results were used for the new interpretation on the entire Korean soils. As this multivariate analysis of soil survey data were attempted for the first time, 5 key factors, which are topography, soil texture(family), drainage class, available soil depth, and gravel content, were selected as classification factors judging from our survey experiences. Based on the results obtained from the analysis of 314 soil series for these 5 factors, Korean soils could be re-classified into 24 groups. The results were designated on 4 quadrants split using the eigenvector for the analysis of similarities of soil properties. Also another interpretation was added by using a method of drawing a circle based on the eigenvalue. Conclusively, re-classification in more simplified ways with a new interpretation by the modification of the past soil survey data was done successfully and it would be helpful for practical use although a few of exceptional soils were distributed in the different soil groups. Homogeneity with the same groups are to be improved by controlling weighting values, or by adding or subtracting some factors in the multivariate analysis. In addition, re-establishment of criteria for soil classification through the more detailed examination are needed.