Wanchang Zhang

Urban building segmentation is a prevalent research domain for very high resolution (VHR) remote sensing; however, various appearances and complicated background of VHR remote sensing imagery make accurate semantic segmentation of urban buildings a challenge in relevant applications. Following the basic architecture of U-Net, an end-to-end deep convolutional neural network (denoted as DeepResUnet) was proposed, which can effectively perform urban building segmentation at pixel scale from VHR imagery and generate accurate segmentation results. The method contains two sub-networks: One is a cascade down-sampling network for extracting feature maps of buildings from the VHR image, and the other is an up-sampling network for reconstructing those extracted feature maps back to the same size of the input VHR image. The deep residual learning approach was adopted to facilitate training in order to alleviate the degradation problem that often occurred in the model training process. The proposed DeepResUnet was tested with aerial images with a spatial resolution of 0.075 m and was compared in performance under the exact same conditions with six other state-of-the-art networks—FCN-8s, SegNet, DeconvNet, U-Net, ResUNet and DeepUNet. Results of extensive experiments indicated that the proposed DeepResUnet outperformed the other six existing networks in semantic segmentation of urban buildings in terms of visual and quantitative evaluation, especially in labeling irregular-shape and small-size buildings with higher accuracy and entirety. Compared with the U-Net, the F1 score, Kappa coefficient and overall accuracy of DeepResUnet were improved by 3.52%, 4.67% and 1.72%, respectively. Moreover, the proposed DeepResUnet required much fewer parameters than the U-Net, highlighting its significant improvement among U-Net applications. Nevertheless, the inference time of DeepResUnet is slightly longer than that of the U-Net, which is subject to further improvement.

In Punjab (Pakistan), the increasing population and expansion of land use for agriculture have severely exploited the regional groundwater resources. Intensive pumping has resulted in a rapid decline in the level of the water table as well as its quality. Better management practices and artificial recharge are needed for the development of sustainable groundwater resources. This study proposes a methodology to delineate favorable groundwater potential recharge zones (FPRI) by integrating maps of groundwater potential recharge index (PRI) with the DRASTIC-based groundwater vulnerability index (VI). In order to evaluate both indexes, different thematic layers corresponding to each index were overlaid in ArcGIS. In the overlay analysis, the weights (for various thematic layers) and rating values (for sub-classes) were allocated based on a review of published literature. Both were then normalized and modified using the analytical hierarchical process (AHP) and a frequency ratio model respectively. After evaluating PRI and FPRI, these maps were validated using the area under the curve (AUC) method. The PRI map indicates that 53% of the area assessed exists in very low to low recharge zones, 22% in moderate, and 25% in high to excellent potential recharge zones. The VI map indicates that 38% of the area assessed exists in very low to low vulnerability, 33% in moderate, and 29% in high to very high vulnerability zones. The FPRI map shows that the central region of Punjab is moderately-to-highly favorable for recharge due to its low vulnerability and high recharge potential. During the validation process, it was found that the AUC estimated with modified weights and rating values was 79% and 67%, for PRI and VI indexes, respectively. The AUC was less when evaluated using original weights and rating values taken from published literature. Maps of favorable groundwater potential recharge zones are helpful for planning and implementation of wells and hydraulic structures in this region.