Surface area is a fundamental parameter derived from terrain analysis in geographical information systems (GIS) for modeling the real world. Therefore, an evaluation of the techniques implemented by GIS models is important in order to obtain reliable area measurements. The study compares two raster-based methods with a TIN-based (triangulated irregular network) method for surface area approximation using a digital elevation model (DEM) with 25 m × 25 m resolution in six mountain regions of Shandong Province, China. Raster-based method Ⅰ calculates surface-area grids from DEMs directly. Eight 3-dimensional triangles connecting the center point of each cell with the center points of the eight surrounding cells are generated and then the areas of the portions of each triangle that lay within the cell boundary are calculated and summed. Raster-based method Ⅱ produces the surface area of each cell directly from the pixel size and the slope value for the pixel. The results demonstrate that the surface areas calculated by raster-based method Ⅰ are equal to those of the TIN-based method. The results of raster-based method Ⅱ are the lowest but the difference in areas between the two raster-based methods decreases with the decrease in terrain complexity. Compared to the method with TINs, raster-based method Ⅱ can do better because it has some advantages, such as neighborhood analysis, more consistent output and faster processing speed. The results also demonstrate that the difference between surface and horizontal areas should be considered if more than 30% of the area of a region has slopes steeper than 18.2 degrees. Therefore, an evaluation of the techniques implemented by GIS models is important in order to obtain reliable area measurements. The study compares two raster- based methods with a TIN-based (triangulated irregular network) method for surface area approximation using a digital elevation model (DEM) with 25 m × 25 m resolution in six mountain regions of Shandong Province, China. Eight 3-dimensional triangles connecting the center point of each cell with the center points of the eight surrounding cells are generated and then the areas of the portions of each triangle that lay within the cell boundary are calculated and summed. Raster-based method II produces the surface area of ​​each cell directly from the pixel size and the slope value for the pixel. hat the surface areas calculated by raster-based method I are equal to those of the TIN-based method. The results of raster-based method II are the lowest but the difference in areas between the two raster-based methods decreases with the decrease in terrain complexity. Compared to the method with TINs, raster-based method II can do better because it has some advantages, such as neighborhood analysis, more consistent output and faster processing speed. The results also demonstrate that the difference between surface and horizontal areas should be considered if more than 30% of the area of ​​a region has slopes steeper than 18.2 degrees.