Spatiotemporal modeling of land subsidence using a geographically weighted deep learning method based on PS-InSAR.

The demand for water resources during urbanization forces the continuous exploitation of groundwater, resulting in dramatic piezometric drawdown and inducing regional land subsidence (LS). This has greatly threatened sustainable development in the long run. LS modeling helps understanding the factors responsible for the ongoing loss of land elevation and hence enhances the development of prevention strategies. Data-driven LS models perform well with fewer variables and faster convergence than physically-based hydrogeological models. However, the former models often cannot simultaneously reflect the temporal nonlinearity and spatial correlation (SC) characteristics of LS under complex variables. We proposed a LS spatiotemporal model which considers both nonlinear and spatial correlations between LS and groundwater level change of exploited aquifers. It is based on deep learning method and LS time series detected by permanent scatterer-interferometric synthetic aperture radar (PS-InSAR). The LS time series and hydrogeological properties are constructed as a spatiotemporal dataset for model training. The spatiotemporal LS model, geographically weighted long short-term memory (GW-LSTM), is constructed by integrating SC with LSTM. This latter is a deep recurrent neural network approach incorporating sequential data. The model is validated by a case study in the Beijing plain. The results show that the accuracy of the proposed model can be greatly improved considering the spatial correlation between LS and influencing factors. Furthermore, the comparison between the LSTM and GW-LSTM models reveals that groundwater level variation is not a unique causation of LS in the study area. The developed model deals with the spatiotemporal characteristics of LS under multiple variables and can be used to predict LS under different scenarios of groundwater level variations for the purpose of monitoring and providing evidence to support the prevention of future LS.

We lose ground: Global assessment of land subsidence impact extent.

Depletion of groundwater aquifers along with all of the associated quality and quantity problems which affect profitability of direct agricultural and urban users and linked groundwater-ecosystems have been recognized globally. During recent years, attention has been devoted to land subsidence-the loss of land elevation that occurs in areas with certain geological characteristics associated with aquifer exploitation. Despite the large socioeconomic impacts of land subsidence most of these effects are still not well analyzed and not properly recognized and quantified globally. In this paper we developed a land subsidence impact extent (LSIE) index that is based on 10 land subsidence attributes, and applied it to 113 sites located around the world with reported land subsidence effects. We used statistical means to map physical, human, and policy variables to the regions affected by land subsidence and quantified their impact on the index. Our main findings suggest that LSIE increases between 0.1 and 6.5% by changes in natural processes, regulatory policy interventions, and groundwater usage, while holding all other variables unchanged. Effectiveness of regulatory policy interventions varies depending on the lithology of the aquifer system, in particular its stiffness. Our findings suggest also that developing countries are more prone to land subsidence due to lower performance of their existing water governance and institutions.

The role of sedimentation and natural compaction in a prograding delta: insights from the mega Mekong delta, Vietnam.

The Vietnamese Mekong Delta was formed by rapid transgression during the second half of the Holocene by deposition of mainly unconsolidated, fine-grained (clayey) sediments undergoing high compaction rates. The natural subsidence can seriously impact the already vulnerable delta plain as its low elevation exposes the delta to global sea level rise, flooding, salinization. Human activities such as groundwater pumping, infrastructural loading, sand mining and dam construction have exacerbated the effects of natural consolidation. Here we present a novel modeling study that has allowed to reproduce the formation and evolution of the Mekong delta over the past 4000 years. Using an adaptive finite-element mesh, the model properly simulates accretion and natural consolidation characterizing the delta evolution. Large soil grain motion and the delayed dissipation of pore-water overpressure are accounted for. We find that natural compaction of Holocene deposits following delta evolution exceeds predicted values of absolute sea level rise. The unprecedented high rates (up to ~20 mm/yr) threaten the lower delta plain with permanent inundation and inevitably reduce the designed service life of flood defense structures along the coast. Total subsidence and sediment delivery to the delta plain will determine its future elevation and vulnerability to relative sea level rise.

The Crotone Megalandslide, southern Italy: Architecture, timing and tectonic control.

Large-scale submarine gravitational land movements involving even more than 1,000 m thick sedimentary successions are known as megalandslides. We prove the existence of large-scale gravitational phenomena off the Crotone Basin, a forearc basin located on the Ionian side of Calabria (southern Italy), by seismic, morpho-bathymetric and well data. Our study reveals that the Crotone Megalandslide started moving between Late Zanclean and Early Piacenzian and was triggered by a contractional tectonic event leading to the basin inversion. Seaward gliding of the megalandslide continued until roughly Late Gelasian, and then resumed since Middle Pleistocene with a modest rate. Interestingly, the onshore part of the basin does not show a gravity-driven deformation comparable to that observed in the marine area, and this peculiar evidence allows some speculations on the origin of the megalandslide.

Saltwater contamination in the managed low-lying farmland of the Venice coast, Italy: An assessment of vulnerability.

The original morphology and hydrogeology of many low-lying coastlands worldwide have been significantly modified over the last century through river diversion, embankment built-up, and large-scale land reclamation projects. This led to a progressive shifting of the groundwater-surficial water exchanges from naturally to anthropogenically driven. In this human-influenced hydrologic landscape, the saltwater contamination usually jeopardizes the soil productivity. In the coastland south of Venice (Italy), several well log measurements, chemical and isotope analyses have been performed over the last decade to characterize the occurrence of the salt contamination. The processing of this huge dataset highlights a permanent variously-shaped saline contamination up to 20km inland, with different conditions in relation with the various geomorphological features of the area. The results point out the important role of the land reclamation in shaping the present-day salt contamination and reveal the contribution of precipitation, river discharge, lagoon and sea water to the shallow groundwater in the various coastal sectors. Moreover, an original vulnerability map to salt contamination in relation to the farmland productivity has been developed taking into account the electrical conductivity of the upper aquifer in the worst condition, the ground elevation, and the distance from salt and fresh surface water sources. Finally, the study allows highlighting the limit of traditional investigations in monitoring saltwater contamination at the regional scale in managed Holocene coastal environments. Possible improvements are outlined.

Natural versus anthropogenic subsidence of Venice.

We detected land displacements of Venice by Persistent Scatterer Interferometry using ERS and ENVISAT C-band and TerraSAR-X and COSMO-SkyMed X-band acquisitions over the periods 1992-2010 and 2008-2011, respectively. By reason of the larger observation period, the C-band sensors was used to quantify the long-term movements, i.e. the subsidence component primarily ascribed to natural processes. The high resolution X-band satellites reveal a high effectiveness to monitor short-time movements as those induced by human activities. Interpolation of the two datasets and removal of the C-band from the X-band map allows discriminating between the natural and anthropogenic components of the subsidence. A certain variability characterizes the natural subsidence (0.9 ± 0.7 mm/yr), mainly because of the heterogeneous nature and age of the lagoon subsoil. The 2008 displacements show that man interventions are responsible for movements ranging from -10 to 2 mm/yr. These displacements are generally local and distributed along the margins of the city islands.

Simultaneous monitoring of soil water content and salinity with a low-cost capacitance-resistance probe.

Capacitance and resistivity sensors can be used to continuously monitor soil volumetric water content (θ) and pore-water electrical conductivity (EC(p)) with non-destructive methods. However, dielectric readings of capacitance sensors operating at low frequencies are normally biased by high soil electrical conductivity. A procedure to calibrate capacitance-resistance probes in saline conditions was implemented in contrasting soils. A low-cost capacitance-resistance probe (ECH2O-5TE, 70 MHz, Decagon Devices, Pullman, WA, USA) was used in five soils at four water contents (i.e., from dry conditions to saturation) and four salinity levels of the wetting solution (0, 5, 10, and 15 dS · m-1). θ was accurately predicted as a function of the dielectric constant, apparent electrical conductivity (EC(a)), texture and organic carbon content, even in high salinity conditions. Four models to estimate pore-water electrical conductivity were tested and a set of empirical predicting functions were identified to estimate the model parameters based on easily available soil properties (e.g., texture, soil organic matter). The four models were reformulated to estimate EC(p) as a function of EC(a), dielectric readings, and soil characteristics, improving their performances with respect to the original model formulation. Low-cost capacitance-resistance probes, if properly calibrated, can be effectively used to monitor water and solute dynamics in saline soils.

Surface flow boundary conditions in modeling land subsidence due to fluid withdrawal.

Land subsidence due to subsurface fluid (water, gas, oil) withdrawal is often predicted by either finite element or finite difference numerical models based on coupled poroelastic theory, where the soil is represented as a semi-infinite medium bounded by the traction-free (ground) surface. One of the variables playing a most important role on the final outcome is the flow condition used on the traction-free boundary, which may be assumed as either permeable or impermeable. Although occasionally justified, the assumption of no-flow surface seems to be in general rather unrealistic. A permeable boundary where the fluid pressure is fixed to the external atmospheric pressure appears to be more appropriate. This paper addresses the response, in terms of land subsidence, obtained with a coupled poroelastic finite element model that simulates a distributed pumping from a horizontal aquifer confined between two relatively impervious layers, and takes either a permeable boundary surface, i.e., constant hydraulic potential, or an impermeable boundary, i.e., a zero Neumann flow condition. The analysis reveals that land subsidence is rather sensitive to the flow condition implemented on the traction-free boundary. In general, the no-flow condition leads to an overestimate of the predicted ground surface settlement, which could even be 1 order of magnitude larger than that obtained with the permeable boundary.