Spatial Distribution Analysis of Ruditapes philippinarum Habitat Using Data Mining

Abstract

Lee, S.M.; Oh, H.J.; Lee, S., and Lee, M.J., 2019. Spatial distribution analysis of Ruditapes philippinarum habitat using data mining. In: Jung, H.-S.; Lee, S.; Ryu, J.-H., and Cui, T. (eds.), Advances in Remote Sensing and Geoscience Information Systems of Coastal Environments. Journal of Coastal Research, Special Issue No. 90, pp. 112-119. Coconut Creek (Florida), ISSN 0749-0208.The purpose of this study is to analyze the spatial distribution of Ruditapes philippinarum habitat in Geunso Bay, South Korea. R. philippinarum samples were acquired through in-situ observation. Remotely sensed data were used to derive factors related to R. philippinarum habitat. A spatial distribution map was generated using a data-mining model focused on the Chi-squared Automatic Interaction Detection (CHAID) model with a spatial dataset of eight R. philippinarum habitat-related factors: training data of R. philippinarum observations, spectral reflectance factor, sediment type factor, tidal channel factors (density of and distance from the tidal channel), and morphological factors (elevation, slope gradient, slope aspect and exposure time). Validation analysis was performed through comparison with observed habitat locations. The accuracy rate of the CHAID model was 84.7 % (area under the curve [AUC] = 0.847) using success rate analysis and 74.5 % (AUC = 0.745) using prediction rate analysis. The data-mining model based on remote sensing data and geographic information system (GIS) tools is an efficient method for mapping the spatial distribution of R. philippinarum habitat in tidal flats.