ALBERT

Description: Long-term excessive groundwater exploitation for agricultural, domestic and stock applications has resulted in substantial ground subsidence in Arizona, USA, and especially in the Willcox Groundwater Basin. The land subsidence rate of the Willcox Basin has not declined but has rather increased in recent years, posing a threat to infrastructure, aquifer systems, and ecological environments. In this study, we first investigate the spatiotemporal characteristics of land subsidence in the Willcox Groundwater Basin using an interferometric synthetic aperture radar (InSAR) time series analytical approach with L-band ALOS and C-band Sentinel-1 SAR data acquired from 2006 to 2020. The overall deformation patterns are characterized by two major zones of subsidence, with the mean subsidence rate increasing with time from 2006 to 2020. An approach based on independent component analysis (ICA) was adopted to separate the mixed InSAR time series signal into a set of independent signals. The application of ICA to the Willcox Basin not only revealed that two different spatiotemporal deformation features exist in the basin but also filtered the residual errors in InSAR observations to enhance the deformation time series. Integrating the InSAR deformation and groundwater level data, the response of the aquifer skeletal system to the change in hydraulic head was quantified, and the hydromechanical properties of the aquifer system were characterized. Historical spatiotemporal storage loss from 1990 to 2020 was also estimated using InSAR measurements, hydraulic head and estimated skeletal storativity. Understanding the characteristics of land surface deformation and quantifying the response of aquifer systems in the Willcox Basin and other groundwater basins elsewhere are important in managing groundwater exploitation to sustain the mechanical health and integrity of aquifer systems.

Description: Ice flow models of the Antarctic ice sheet are commonly used to simulate its future evolution in response to different climate scenarios and assess the mass loss that would contribute to future sea level rise. However, there is currently no consensus on estimates of the future mass balance of the ice sheet, primarily because of differences in the representation of physical processes, forcings employed and initial states of ice sheet models. This study presents results from ice flow model simulations from 13 international groups focusing on the evolution of the Antarctic ice sheet during the period 2015–2100 as part of the Ice Sheet Model Intercomparison for CMIP6 (ISMIP6). They are forced with outputs from a subset of models from the Coupled Model Intercomparison Project Phase 5 (CMIP5), representative of the spread in climate model results. Simulations of the Antarctic ice sheet contribution to sea level rise in response to increased warming during this period varies between −7.8 and 30.0 cm of sea level equivalent (SLE) under Representative Concentration Pathway (RCP) 8.5 scenario forcing. These numbers are relative to a control experiment with constant climate conditions and should therefore be added to the mass loss contribution under climate conditions similar to present-day conditions over the same period. The simulated evolution of the West Antarctic ice sheet varies widely among models, with an overall mass loss, up to 18.0 cm SLE, in response to changes in oceanic conditions. East Antarctica mass change varies between −6.1 and 8.3 cm SLE in the simulations, with a significant increase in surface mass balance outweighing the increased ice discharge under most RCP 8.5 scenario forcings. The inclusion of ice shelf collapse, here assumed to be caused by large amounts of liquid water ponding at the surface of ice shelves, yields an additional simulated mass loss of 28 mm compared to simulations without ice shelf collapse. The largest sources of uncertainty come from the climate forcing, the ocean-induced melt rates, the calibration of these melt rates based on oceanic conditions taken outside of ice shelf cavities and the ice sheet dynamic response to these oceanic changes. Results under RCP 2.6 scenario based on two CMIP5 climate models show an additional mass loss of 0 and 3 cm of SLE on average compared to simulations done under present-day conditions for the two CMIP5 forcings used and display limited mass gain in East Antarctica.

Description: Semantic segmentation and depth completion are two challenging tasks in scene understanding, and they are widely used in robotics and autonomous driving. Although several studies have been proposed to jointly train these two tasks using some small modifications, such as changing the last layer, the result of one task is not utilized to improve the performance of the other one despite that there are some similarities between these two tasks. In this article, we propose multitask generative adversarial networks (Multitask GANs), which are not only competent in semantic segmentation and depth completion but also improve the accuracy of depth completion through generated semantic images. In addition, we improve the details of generated semantic images based on CycleGAN by introducing multiscale spatial pooling blocks and the structural similarity reconstruction loss. Furthermore, considering the inner consistency between semantic and geometric structures, we develop a semantic-guided smoothness loss to improve depth completion results. Extensive experiments on the Cityscapes data set and the KITTI depth completion benchmark show that the Multitask GANs are capable of achieving competitive performance for both semantic segmentation and depth completion tasks.

Description: This archive provides the scripts and routines used as part of the publication "ISMIP6 Antarctica: a multi-model ensemble of the Antarctic ice sheet evolution over the 21st century", published in The Cryosphere, https://tc.copernicus.org/articles/14/3033/2020/

Description: Deep learning-based methods have achieved remarkable performance in 3-D sensing since they perceive environments in a biologically inspired manner. Nevertheless, the existing approaches trained by monocular sequences are still prone to fail in dynamic environments. In this work, we mitigate the negative influence of dynamic environments on the joint estimation of depth and visual odometry (VO) through hybrid masks. Since both the VO estimation and view reconstruction process in the joint estimation framework is vulnerable to dynamic environments, we propose the cover mask and the filter mask to alleviate the adverse effects, respectively. As the depth and VO estimation are tightly coupled during training, the improved VO estimation promotes depth estimation as well. Besides, a depth-pose consistency loss is proposed to overcome the scale inconsistency between different training samples of monocular sequences. Experimental results show that both our depth prediction and globally consistent VO estimation are state of the art when evaluated on the KITTI benchmark. We evaluate our depth prediction model on the Make3D dataset to prove the transferability of our method as well.