Wetlands are an important natural resource that provide many benefits to society (e.g. water storage, flood mitigation) and provide critical habitat to plants and animals. The timing and duration of flooding within a wetland largely determines the ecosystem services and the species a wetland supports. However, landscape-level hydrologic data for wetlands is scarce because tracking changes in wetland water levels over weeks and months requires the installation of expensive monitoring equipment or visiting sites many times a year for several years. Remote sensing is a useful tool to understand wetland dynamics. Synthetic Aperture Radar (SAR) is an especially useful tool because of the all-weather capability. Each pixel in a SAR image stores backscattering intensity that reflects geospatial characteristics (e.g. surface roughness, soil moisture) of its corresponding area. Because of specular reflection, water surfaces have lower backscattering intensity than land surfaces. Many researchers have proposed classification models considering this phenomenon. However, previous studies reported that a large local incidence angle of microwave also decreases backscattering intensity of land surfaces. To mitigate the misclassification derived from this effect, this study proposes multi-incidence angle SAR image analysis, which uses two images that are taken from the ascending and descending path. This model utilizes a two-dimensional space defined by intensity values stored in each pixel, and separates water pixels and land pixels. The classification ability of the model was tested on two Sentinel-1 images capturing wetlands in Okanogan County, Washington. This study successfully demonstrates that combining two images can detect water surfaces more accurately than the previous model that considers only one image.