Data island effectively blocks the practical application of machine learning. To meet this challenge, a new framework known as federated learning was created. It allows model training on a large amount of scattered data owned by different data providers. This article presents a parallel solution for computing logistic regression based on distributed asynchronous task framework. Compared to the existing work, our proposed solution does not rely on any third-party coordinator, and hence has better security and can solve the multitraining problem. The logistic regression based on homomorphic encryption is implemented in Python, which is used for vertical federated learning and prediction of the resulting model. We evaluate the proposed solution using the MNIST dataset, and the experimental results show that good performance is achieved.

The risk of coastal storm flooding is deteriorating under global warming, especially for the heavily urbanized deltaic cities, like Shanghai. The Nature-Based Flood Defense (NBFD), as an eco-friendly design alternative for hard infrastructure against coastal flooding, is gaining attention. Nevertheless, the vulnerability of saltmarsh due to the biological instability, resulting in the uncertainties on coastal protection, is considered the bottleneck challenge that hinders the broad application of the NBFD concept. We argue that except for direct wave attenuations by the above-ground vegetation during storms, the gradual sediment trapping and consolidating during the non-storm period is a more crucial function of coastal saltmarsh, which mitigates storm waves by forming a broader and higher intertidal morphology. This benefit is an important value of saltmarsh-based coastal protection but is largely neglected in many NBFD studies. Taking Chongming Dongtan Shoal (CDS) as a case study, we demonstrated that over 2/3th wave attenuation during storms is contributed by the saltmarsh morphology, and less than 1/3th is from the saltmarsh vegetation. The relative contribution of the saltmarsh morphology on wave mitigation is even enhanced under the increasing storm grades from 100 yrs. to 5000 yrs. return levels. To promote this idea for broader application, the cost-benefit analysis of three artificial NBFD solutions (e.g., submerged breakwater, timber piles, and sand nourishment) are compared. We identified an optimal measure of the submerged breakwater for CDS, which minimizes the ecological impact and maximizes the cost-benefit. Moreover, the wave-free zone behind the breakwater increases the chance of vegetation establishment, helps suspended sediment trapping, hence fostering a beneficent cycle for saltmarsh restoration. In summary, ignoring the contribution of saltmarsh morphology on wave attenuation largely underestimated the benefits of vegetation-based coastal protection, which should be greatly emphasized to provide a solid basis for developing NBFD.

A better understanding of how tidal-flat reclamation changes the flood hazard is critical for climate-proofing coastal flood defense design of heavily urbanized areas. Since the 1950s, large-scale reclamation has been performed along the Shanghai coast, China, to fulfill the land demands of city expansion. We now show that the loss of tidal flats may have resulted in harmful impacts of coastal storm flooding. Using the foreshore profiles measured before and after reclamation (i.e., wide vs. narrow tidal flat), we determined the long-term changes in flood risk using a numerical model that combines extreme tidal level and wave overtopping analysis. Results show that wide tidal flats in front of a seawall provide efficient wave damping even during extreme water levels. Reclamation of these tidal flats substantially increased wave heights and correspondingly reduced the return period of a specific storm. As a result, estimates of overtopping are aggravated by more than 80% for the varying return periods examined. It is concluded that the disasters of coastal flooding after the 1997 tidal-flat reclamation in Hangzhou Bay, China are a consequence of both anthropogenic and natural activities. Moreover, our model calculations provide an equation describing the equivalent dike height needed to compensate for the loss of every km tidal flat of a certain elevation, and vice versa. For example, for every km of tidal flat ranging from high marsh to bare tidal flat that is being regained, the dike can be lowered by 0.84 m–0.67 m, when designing for a 1 in 200 years storm event. Overall, we suggest that wide tidal flats are ideally restored in front of dikes, and that when tidal areas are reclaimed, the seawall height is raised as part of the intertidal reclamation procedure. Using such an equivalent protection standard is relevant to designing hybrid flood defense system worldwide.

Flood risk is expected to increase in coastal cities, particularly in Asian megacities such as Shanghai. This paper presents an integrated modeling framework to simulate changes in the flood risk in Shanghai and provide a cost-benefit analysis of multiple adaptation strategies used to reduce risk. The results show that the potential flood risk will increase dramatically as a result of sea level rise, land subsidence, and socioeconomic development. By 2100, the expected annual damage could reach 0.8% (uncertainty range: 0.4%–1.4%) of local GDP under an optimistic emission scenario (RCP4.5), compared to the current value of 0.03%. All of the adaptation strategies can effectively reduce the flood risk under the current conditions and those in 2050. In contrast to the ‘hard’ flood protection strategies (i.e., storm-surge barriers and floodwalls), the ‘soft’ strategies (i.e., building codes and nature-based measures) cannot substantially reduce the flood risk in 2100. However, the soft strategies can play a critical role in reducing the residual risk resulting from the hard strategies. A ‘hybrid’ strategy combining a storm-surge barrier, wet-proofing, and coastal wetland development outperforms both hard and soft strategies in terms of low residual risk and high benefit/cost ratio. Additionally, the hybrid strategy can also enable a larger reduction in casualties. These findings imply that managing flood risk is more than the use of single adaptation measures. The methodology developed in this paper can enlighten Shanghai and other coastal cities on an economically and socially feasible adaptation strategy in an uncertain future.

Extreme flooding usually causes huge losses of residential buildings and household properties, which is critical to flood risk analysis and flood resilience building in Shanghai. We developed a scenario-based multidisciplinary approach to analyze the exposure, losses and risks of residential buildings and household properties, and their spatial patterns at the neighborhood committee level in Shanghai, based on extreme storm flood scenarios of 1/200, 1/500, 1/1000 and 1/5000-year. Our findings show that the inundation area of the residential buildings caused by a 1/200-year storm flood reaches 24.9 km², and the total loss of residential buildings and household properties is 29.7 billion CNY (Chinese Yuan) (or 4.4 billion USD), while the inundation area of residential buildings and the total loss increases up to 162.4 km² and 366.0 billion CNY (or 54.2 billion USD), respectively for a 1/5000-year storm flood. The estimated average annual loss (AAL) of residential buildings and household properties for Shanghai is 590 million CNY/year (or 87.4 million USD/year), with several hot spots distributed around the main urban area and on the bank of the Hangzhou Bay. Among sixteen districts, Pudong has the highest exposure and annual expected loss, while the inner city is also subject to extreme flooding with an AAL up to near half of the total. An analysis of flood risk in each of 209 subdistricts/towns finds that those most vulnerable to storm flooding are concentrated in Pudong, Jiading, Baoshan Districts and the inner city. Our work can provide meaningful information for risk-sensitive urban planning and resilience building in Shanghai. The methodology can also be used for risk analysis in other coastal cities facing the threat of storm flooding.

The geomorphology of the Yangtze Estuary in the Changjiang River Delta in Eastern China has been the subject of extensive research. This study extends previous work to examine the influence of wind-waves on the mouth-bar, where about half of the river-borne material settles to the bed. The site is located just outside of Changjiang River mouth, which is meso-tidal and subject to seasonally varying river flows and wind-wave conditions. Modeling was performed with a coupled wave-current hydrodynamic model using TELEMAC and TOMAWAC and validated against observed data. Bottom Shear Stress (BSS) from river, tide and waves based on the numerical model output was used to infer the respective contribution to the evolution of the subaqueous delta. Our examination did not however extend to modeling the sediment transport or the morphological bed changes. The results suggest that (i) the dominance of river discharge is limited to an area inside the mouth, while outside, the mouth-bar is tide-wave dominant; (ii) considering just the tide, the currents on the shallow shoals are flood dominant and deep channels are ebb dominant, which induces continued accretion over the shallows and erodes the deeper parts of the mouth-bar until the tidal currents become too weak to transport sediment; (iii) whereas waves are very efficient at reshaping the shallow shoals, with the effect being subtly dependent on the depth distribution over the mouth-bar; (iv) the stability of shallow shoal morphology is highly dependent on the presence of seasonal wind-waves and characterized as “summer storing and winter erosion” while deep channels perform like corridors of water and sediment, exporting sediment all year round. The nature of the mouth-bar response has important implications for coastal management, such as the ongoing deep water channel maintenance, reclamations and coastal defense measures.