Impervious surface mapping is of great importance in urban studies. Impervious surfaces are major components of urban infrastructures and their expansion represents urban development. These surfaces mainly include built-up areas and streets; they are composed of various materials and found in diverse sizes and shapes. Impervious surface detection is challenging due to the confusion of these surfaces with other land cover classes. These confusions are not constant over different seasons, as seasonality affects the target’s responses. This study particularly focused on the seasonal effect on impervious surface detection using Sentinel-1 and Sentinel-2 images to find the optimum season. The study area is the city of Sanandaj, in the west of Iran. All processes have been executed on the Google Earth Engine as it provides a platform to access and process the satellite images. To exclude the effect of the classification algorithm on the obtained results, three commonly used classifiers have been compared; i.e., maximum likelihood, support vector machine, and neural network. The results show that spring is the best season to delineate impervious surfaces from remaining land covers, while the use of winter images does not provide acceptable results. Sentinel-2 results outperform Sentinel-1. Variation in topography and high sensitivity of SAR responses to moisture and volume structure hinder the application of Sentinel-1 images in the heterogeneous urban area. The built-up class has higher producer accuracy as compared to the street class. There was considerable confusion between the street and bare soil classes in both Sentinel-1 and Sentinel-2 images.

不透水表面测绘在城市研究中非常重要。不透水表面是城市基础设施的主要组成部分，其扩张代表了城市发展。这些表面主要包括建成区和街道；它们由各种材料组成，大小和形状各异。由于这些表面与其他土地覆盖类别的混淆，不透水表面检测具有挑战性。由于季节性会影响目标的响应，因此这些混淆在不同的季节并不一致。本研究特别关注使用 Sentinel-1 和 Sentinel-2 图像寻找最佳季节对不透水表面检测的季节性影响。研究区域是伊朗西部的萨南达季市。所有过程都在 Google Earth Engine 上执行，因为它提供了一个访问和处理卫星图像的平台。为了排除分类算法对所得结果的影响，比较了三种常用的分类器；即最大似然、支持向量机和神经网络。结果表明，春季是根据剩余土地覆盖物描绘不透水表面的最佳季节，而使用冬季图像无法提供可接受的结果。Sentinel-2 结果优于 Sentinel-1。地形的变化和 SAR 对水分和体积结构响应的高灵敏度阻碍了 Sentinel-1 图像在异构城市地区的应用。与街道类相比，建成类具有更高的生产者准确度。