This paper generated time-series temperature maps and analyzed the characteristics of temperature distributions from monthly average temperature observations between 2010 and 2011 in Jirisan areas using topographic data and geostatistics. From variogram modeling, all months except May to August showed that the spatial variability of temperature was the greatest along the direction perpendicular to coasts. Monthly temperature has negative correlations with elevation and distances from coasts and especially the correlation between temperature and distances from coasts was very weak in summer like the variogram modeling result. For temperature distribution mapping, kriging with a trend and ordinary kriging were separately applied as a univariate kriging algorithm by considering the spatial variability structures of temperature. Simple kriging with varying local means was applied as a multivariate kriging algorithm for integrating topographic data sets. From the cross validation results, the use of topographic data in spatial prediction of temperature showed the improved predictive performance, compared with univariate kriging. This improvement in predictive performance was dependent mainly on mean and variation values of monthly temperature and the spatial auto-correlation strength of residuals, as well as the correlation between topographic data and temperature. Based on these analysis results, spatial variability analysis using variogram is effectively used to account for spatial characteristics of monthly temperature and the correlation with topographic data. Topographic data can also be a useful information source for reliable temperature mapping.

• 박노욱(인하대학교 지리정보공학과) | No-Wook Park
• 장동호(공주대학교 지리학과) | Dong-Ho Jang Correspondence