*2.2. Materials*

We used three types of data for our analyses: remote sensing (RS), geographic and statistical (Table 1). The RS data consisted of 2010 and 2016 land cover maps produced by applying maximum-likelihood supervised classification to medium-resolution satellite imagery using Envi 5.0 software. The land cover map was made using three scenes (path 116/row 33, path 116/row 34, path 115/row 33) and using Landsat 5 and Landsat 8 images. The classification accuracy of the land cover map was as follows, Overall accuracy: 93, Kappa: 0.91 in 2010, Overall accuracy: 943, Kappa: 0.92. The classification accuracy of the land cover maps was high in both overall accuracy and Kappa. Reference data for the accuracy of classification maps were used for field survey and Google earth images. The geographic data consisted of geographic information system (GIS) layers for administrative boundaries map provided by the Korea Forestry Service, the Southern Limit Line provided by the Gangwon Development Research Institute and CCZ, Digital topographic maps (scale of 1:25,000), cadastral valuation data and local population data provided by the Ministry of Land, Infrastructure and Transport. The statistical data consisted of tourist attraction information provided by the Korea Tourism Organization as spatial information. Typical tourist attractions of CCZ are "Unification Observatory", "DMZ Museum" and "Dam of Peace".


**Table 1.** Data used for analyzing the forest LULC patterns and change factors.
