Landsat imagery is an unparalleled freely available data source that allows reconstructing land-cover and land-use change, including horizontal and vertical urban form. This paper addresses the challenge of using Landsat data, particularly its 30m spatial resolution, for monitoring three-dimensional urban densification. Unlike conventional convolutional neural networks (CNNs) for scene recognition resulting in resolution loss, the proposed semantic segmentation framework provides a pixel-wise classification and improves the accuracy of urban form mapping. We compare temporal and spatial transferability of an adapted DeepLab model with a simple fully convolutional network (FCN) and a texture-based random forest (RF) model to map urban density in the two morphological dimensions: horizontal (compact, open, sparse) and vertical (high rise, low rise). We test whether a model trained on the 2014 data can be applied to 2006 and 1995 for Denmark, and examine whether we could use the model trained on the Danish data to accurately map ten other European cities. Our results show that an implementation of deep networks and the inclusion of multi-scale contextual information greatly improve the classification and the model's ability to generalize across space and time. Between the two semantic segmentation models, DeepLab provides more accurate horizontal and vertical classifications than FCN when sufficient training data is available. By using DeepLab, the F1 score can be increased by 4 and 10 percentage points for detecting vertical urban growth compared to FCN and RF for Denmark. For mapping the ten other European cities with training data from Denmark, DeepLab also shows an advantage of 6 percentage points over RF for both horizontal and vertical dimensions. The resulting maps across the years 1985 to 2018 reveal different patterns of urban growth between Copenhagen and Aarhus, the two largest cities in Denmark, illustrating that those cities have used various planning policies in addressing population growth and housing supply challenges. In summary, we propose a transferable deep learning approach for automated, long-term mapping of urban form from Landsat images that is effective in areas experiencing a slow pace of urban growth or with small-scale changes.

Monitoring long-term landslide activity is of importance for risk assessment and land management. Daytime airborne drones or very high-resolution optical satellites are often used to create landslide maps. However, such imagery comes at a high cost, making long-term risk analysis cost-prohibitive. Despite the widespread use of open-access 30m Landsat imagery, their utility for landslide detection is often limited due to low classification accuracy. One of the major challenges is to separate landslides from other anthropogenic disturbances. Here, we produce landslide maps retrospectively from 1998 to 2017 for landslide-prone and highly populated Taiwan (35,874 km 2). To improve classification accuracy of landslides, we integrate nighttime light imagery from the Defense Meteorological Satellite Program (DMSP) and the Visible Infrared Imaging Radiometer Suite (VIIRS), with multi-seasonal daytime optical Landsat time-series, and digital elevation data from the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER). We employed a non-parametric machine-learning classifier, random forest, to classify the satellite imagery. The classifier was trained with data from three years (2005, 2010, and 2015), and was validated with an independent reference sample from twelve years. Our results demonstrated that combining nighttime light data and multi-seasonal imagery significantly improved the classification (p<0.001), compared to conventional methods based on single-season optical imagery. The results confirmed that the developed classification model enabled mapping of landslides across Taiwan over a long period with annual overall accuracy varying between 96% and 97%, user's and producer's accuracies between 73% and 86%. Spatiotemporal analysis of the landslide inventories from 1998 to 2017 revealed different temporal patterns of landslide activities, showing those areas where landslides were persistent and other areas where landslides tended to reoccur after vegetation regrowth. In sum, we provide a robust method to detect long-term landslide activities based on freely available satellite imagery, which can be applied elsewhere. Our mapping effort of landslide spatiotemporal patterns is expected to be of high importance in developing effective landslide remediation strategies.

This article was submitted to ISPRS Journal of Photogrammetry and Remote Sensing.Human settlement extent (HSE) information is a valuable indicator of worldwide urbanization as well as the resulting human pressure on the natural environment. Therefore, mapping HSE is critical for various environmental issues at local, regional, and even global scales. This paper presents a deep-learning-based framework to automatically map HSE from multi-spectral Sentinel-2 data using regionally available geo-products as training labels. A straightforward, simple, yet effective fully convolutional network-based architecture, Sen2HSE, is implemented as an example for semantic segmentation within the framework. The framework is validated against both manually labelled checking points distributed evenly over the test areas, and * the OpenStreetMap building layer. The HSE mapping results were extensively compared to several baseline products in order to thoroughly evaluate the effectiveness of the proposed HSE mapping framework. The HSE mapping power is consistently demonstrated over 10 representative areas across the world. We also present one regional-scale and one country-wide HSE mapping example from our framework to show the potential for upscaling. The results of this study contribute to the generalization of the applicability of CNN-based approaches for large-scale urban mapping to cases where no up-to-date and accurate ground truth is available, as well as the subsequent monitor of global urbanization.