Time-lapse Gravity Research Articles

Stochastic control of geological carbon storage operations using geophysical monitoring and deep reinforcement learning

Geological carbon storage (GCS) is the process of injecting and storing carbon dioxide (CO2) in the subsurface to reduce greenhouse gas emissions. Safe and profitable GCS operations require effective decision-making in the presence of uncertain geological models, a process which can often be facilitated with geophysical monitoring. In this study, we examine how sequential decision-making algorithms can be combined with geophysical measurements for the optimal control of GCS operations. Specifically, we develop an artificial intelligence model using deep reinforcement learning (DRL) that takes geophysical time-lapse gravity and well pressure monitoring data as input and delivers an optimal CO2 injection policy. The objective of the problem at hand is to maximize the profit of a hypothetical GCS operation while mitigating the potential for induced seismicity, by training DRL agents using combined geostatistical, reservoir and geophysical simulation. Comparisons against two benchmarks – a constant injection strategy and an injection schedule optimized using a commercial reservoir simulator toolbox – show that the stochastic control of such operations from subsurface monitoring data using deep reinforcement learning is feasible. Evaluation results show that DRL agent behavior generates profits which are on average 1 to 8 percent higher than what is possible through a constant injection approach. Furthermore, we show that DRL can generate optimal injection policies applicable to the true (yet previously unseen) subsurface given carefully managed levels of uncertainty.

Vertical gravity gradient in volcano monitoring – In situ measured or theoretical? (Campi Flegrei study)

We analyse the vertical gravity gradient (VGG) properties at calderas using the Campi Flegrei (CF) site in Italy. In situ observed VGG values can depart significantly from the theoretical (normal) value of −308.6 μGal/m, particularly in areas of rugged relief. It is assumed that in sufficiently flat areas, the effect of geology, i.e., of the subsurface density heterogeneities, on VGG could prevail over the effect of terrain (topography), which can subsequently be neglected. With respect to the CF caldera, which is often considered as ‘reasonably flat area’, according to our findings the effect of topography on VGG is usually underestimated, while the effect of deeper geology is overestimated. We model the effect of the near topography on VGG at CF and subsequently verify the results of modelling by in situ observations to support our predictions. The results show that, in terms of VGG, the topographic relief plays a more significant role than the assumed geological sources even at ‘flat’ calderas such as CF. For a better understanding, in addition to CF, we analyse the effect of deeper geological sources on VGG also in the territory of Slovakia using a detailed gravimetric database of Slovakia. As a result, we question the use of in situ observed VGG values when processing and interpreting observed time-lapse gravity changes in volcanic areas accompanied by surface deformation.

Turunan Radial dari Anomali Gravitasi untuk Identifikasi Injeksi Air pada Reservoir

Time Lapse micro-gravity or 4D microgravity has been developed and applied for various purposes related to the dynamics change beneath the earth's surface. The principle is a repeated and periodic measurement to determine the change in gravity value at time interval as a yield of the dynamics changes in the density change of the subsurface. This method can be used for monitoring and identify water or gas injections in reservoirs. Mass density taht changes due to gas injection provide changes in the values of gravity over time. This change allows us to identify the dynamics of gas movement in reservoir. A reservoir model has been made in this research with gas injected in reservoir bulk. The time lapse gravity respons value is calculated numerically. Oil-gas contact is an edge between gas injection front with oil which move direction away from injection point. The first and second radial derivatives of 4D gravity can provide as a front of the gas movement in the reservoir. A radial derivatives is a horizontal derivative of gravity anomaly along radial direction from a point which selected as a entre of the anomaly. The second radial derivative is derivative of radial deriative. Radial derivatives value of this model response have range of 20 microGals/m and second radial derivative has a values of 0.05 microGals/m2. Graphics of absolute value of radial derivative and second radial derivative of 4D gravity anomaly seemingly correlated with the oil-gas contact of reservoir models. Radial derivative and second radial derivative has been proven as new available method wich has ability to shows a boundary of a horizontal layer with vertical density contrast. Keywords: Radial derivatives, reservoir, gas injection.

Time-Lapse Gravity and Subsidence Applied in History Matching of a Gas-Condensate Field

Time-Lapse Gravity and Subsidence Applied in History Matching of a Gas-Condensate Field

Hydrological and volcano-related gravity signals at Mt. Somma–Vesuvius from ∼20 yr of time-lapse gravity monitoring: implications for volcano quiescence

SUMMARY We report on about 20 yr of relative gravity measurements, acquired on Mt. Somma–Vesuvius volcano in order to investigate the hydrological and volcano-tectonic processes controlling the present-day activity of the volcano. The retrieved long-term field of time gravity change (2003–2022) shows a pattern essentially related to the subsidence, which have affected the central part of the volcano, as detected by the permanent GNSS network and InSAR data. After reducing the observations for the effect of vertical deformation, no significant residuals are found, indicating no significant mass accumulation or loss within the volcanic system. In the north-western sector of the study area, at the border of the volcano edifice, however, significant residual positive gravity changes are detected which are associated to ground-water rebound after years of intense exploitation of the aquifers. On the seasonal timescale, we find that stations within the caldera rim are affected by the seasonal hydrological effects, while the gravity stations at the base of the Vesuvius show a less clear correlation. Furthermore, within the caldera rim a multiyear gravity transient is detected with an increase phase lasting about 4 yr followed by a slower decrease phase. Analysis of rain data seem to exclude a hydrological origin, hence, we hypothesize a deeper source related to the geothermal activity, which can be present even if the volcano is in a quiescent state. We infer the depth and volume of the source by inverting the spatial pattern of the gravity field at the peak of the transient. A volume of fluids of 9.5 × 107 m3 with density of 1000 kg m−3 at 2.3 km depth is capable to fit reasonably well the observations. To explain the gravity transient, simple synthetic models are produced, that simulate the ascent of fluids from a deep reservoir up to the depth of 2.3 km and a successive diffusion within the carbonate aquifer hosting the geothermal system. The whole process appears to not significantly affect the seismicity rate and the deformation of the volcano. This study demonstrates the importance of a 4-D gravity monitoring of a volcano to understand its complex gravity signals that cover different spatial and temporal scales. Discriminating the different contributions that mix up in the observed gravity changes, in particular those due to hydrologic/anthropogenic activities form those due to the geothermal dynamics, is fundamental for a complete and reliable evaluation of the volcano state.

GTOOLS, an open-source MATLAB program for processing high precision, relative gravity data for time-lapse gravity monitoring

gTOOLS is an open-source software for the processing of relative gravity data. gTOOLS is available in MATLAB and as a compiled executable to be run under the free MATLAB Runtime Compiler. The software has been designed for time-lapse (temporal) gravity monitoring. Although programmed to read the Scintrex CG-5 and CG-6 gravimeters output data files, it can be easily modified to read data files from other gravimeters. The software binds together single-task processing modules within a very simple user interface that is based on one text file. Gravity processing involves three modules: (a) gravimeter calibration; (b) automatic processing of gravity data to find adjusted gravity differences; and (c) post processing of results. Each module is optional and runs independently from the others. Data processing includes (a) averaging out the measurements noise, and correction for solid Earth tides, and ocean loading, and residual instrumental drift, and (b) calculate the residual instrumental drift and gravity differences between the base station and monitoring sites, and their uncertainties, by a weighted least square analysis of the gravity data. The software allows the automatic processing of a gravity campaign spanning multiple days in a single run. The software is tested on gravity data from 2015 eruption at Cotopaxi volcano, Ecuador.

Hybrid Gravimetry to Map Water Storage Dynamics in a Mountain Catchment

In mountain areas, both the ecosystem and the local population highly depend on water availability. However, water storage dynamics in mountains is challenging to assess because it is highly variable both in time and space. This calls for innovative observation methods that can tackle such measurement challenge. Among them, gravimetry is particularly well-suited as it is directly sensitive–in the sense it does not require any petrophysical relationship–to temporal changes in water content occurring at surface or underground at an intermediate spatial scale (i.e., in a radius of 100 m). To provide constrains on water storage changes in a small headwater catchment (Strengbach catchment, France), we implemented a hybrid gravity approach combining in-situ precise continuous gravity monitoring using a superconducting gravimeter, with relative time-lapse gravity made with a portable Scintrex CG5 gravimeter over a network of 16 stations. This paper presents the resulting spatio-temporal changes in gravity and discusses them in terms of spatial heterogeneities of water storage. We interpret the spatio-temporal changes in gravity by means of: (i) a topography model which assumes spatially homogeneous water storage changes within the catchment, (ii) the topographic wetness index, and (iii) for the first time to our knowledge in a mountain context, by means of a physically based distributed hydrological model. This study therefore demonstrates the ability of hybrid gravimetry to assess the water storage dynamics in a mountain hydrosystem and shows that it provides observations not presumed by the applied physically based distributed hydrological model.

Interpretation of spatiotemporal gravity changes accompanying the earthquake of 21 August 2017 on Ischia (Italy)

We analyse spatiotemporal gravity changes observed on the Ischia island (Italy) accompanying the destructive earthquake of 21 August 2017. The 29 May 2016 to 22 September 2017 time-lapse gravity changes observed at 18 benchmarks of the Ischia gravimetric network are first corrected for the gravitational effect of the surface deformation using the deformation-induced topographic effect (DITE) correction. The co-seismic DITE is computed by Newtonian volumetric integration using the Toposk software, a high-resolution LiDAR DEM and the co-seismic vertical displacement field derived from Sentinel-1 InSAR data. We compare numerically the DITE field with its commonly used Bouguer approximation over the island of Ischia with the outcome that the Bouguer approximation of DITE is adequate and accurate in this case. The residual gravity changes are then computed at gravity benchmarks by correcting the observed gravity changes for the planar Bouguer effect of the elevation changes at benchmarks over the same period. The residual gravity changes are then inverted using an inversion approach based on model exploration and growing source bodies, making use of the Growth-dg inversion tool. The found inversion model, given as subsurface time-lapse density changes, is then interpreted as mainly due to a co-seismic or post-seismic disturbance of the hydrothermal system of the island. Pros and weak points of such interpretation are discussed.

Time-lapse borehole gravity imaging of CO2 injection and withdrawal in a closed carbonate reef

Three borehole gravity (BHG) surveys were performed in 2013, 2016, and 2018 to monitor the changes in gravity/density as a result of the injection and withdrawal of carbon dioxide ([Formula: see text]) into and out of the Dover 33 carbonate reservoir reef in Northern Michigan. The observed gravity changes and inferred density changes have been modeled to determine the flow and storage zones of the injected [Formula: see text] in the reef. The high quality and low level of uncertainty of the data collected make them useful for delineating the [Formula: see text] plume position over time and for identifying the oil sweeping extent and mechanisms in the Dover 33 reef. The time-lapse gravity results indicate the effects of the changing [Formula: see text] mass within the reservoir, consistent with increasing mass from 2013 to 2016 (following [Formula: see text] injection) and a decreasing mass from 2016 to 2018 (after [Formula: see text] withdrawal). Three-dimensional imaging of fluid migrations in the reef has been obtained by coupling the time-lapse BHG results to a 3D porosity and permeability model. This coupled approach allows the evaluation of the volume of the reef affected by the injection of [Formula: see text] between 2013 and 2016, the efficiency of the oil sweeping between 2016 and 2018, and the location of the residual [Formula: see text] plume in the reef after 2018.

Variability of the water stock dynamics in karst: insights from surface-to-tunnel geophysics

A hydrogeophysical field experiment was conducted on a karst hydrosystem in the south of France to investigate groundwater transfer and storage variability at a scale of a few hundred meters. A 200-m-long N/S tunnel going through limestone provided the unique opportunity to set up measurements with original configurations inside the unsaturated zone. Three geophysical methods were used: gravimetry, electrical, and seismic. Two-dimensional electrical resistivity and seismic velocity images were retrieved by surrounding the medium with electrodes and geophones, both at the surface and inside the tunnel to improve the sensitivity in depth. This gave information about the weathering state but also about the limestone content and associated porosity characteristics, as the methods are sensitive to distinct properties with different resolution patterns. A time-lapse gravity surface-to-tunnel profile supplied information on the seasonal water mass changes and its variations along the tunnel. Besides, tracers were injected on each side of the profile from the surface and the restitution was sampled in the tunnel drip flows. A contrasting hydrological behavior was evidenced on each side of the tunnel from temporal gravity measurements and tracing tests. The analysis of the whole dataset allowed for better interpretation of the imaged structures, with different hydrological functioning. This study demonstrates the variability of the karst behavior at the scale of a few hundred meters and the benefits of a multi-method approach coupling hydrological and geophysical measurements. This kind of experiment provides fundamental understanding of systems that cannot be directly observed.

Study Investigates Using Borehole Gravity To Improve Reservoir Monitoring

This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper OTC 30626, “Brazil Presalt, Santos Basin: Feasibility Study for the Application of Borehole Gravity To Improve Reservoir Monitoring,” by Zhijun Du, Adrian Topham, SPE, and Jeremy C. Lofts, SPE, Silicon Microgravity, et al., prepared for the 2020 Offshore Technology Conference, originally scheduled to be held in Houston, 4-7 May. The paper has not been peer reviewed. Copyright 2020 Offshore Technology Conference. Reproduced by permission. The complete paper uses a feasibility study to present the potential application of a three-axis gravimeter borehole measurement in the Libra presalt reservoir in the Santos Basin offshore Brazil. The authors’ findings suggest that an annual survey with a limited well stock could be effective in monitoring this type of reservoir and that a wireline-deployed three-axis gravity tool is likely to provide significant additional surveillance to constrain a reservoir production strategy through better appreciation of the direction of water movement. Introduction The giant Brazilian presalt reservoirs are in ultradeep water (greater than 2000 m) and buried at a depth surpassing 5000 m. In order to conduct reservoir monitoring and achieve successful reservoir management, borehole-based geophysical technologies are preferred because the measurements can be made closer to the reservoir mass. The development of a borehole gravimeter suit-able for use in deep reservoirs, however, poses significant technical challenges, and breakthroughs have been limited by the sensor form factor (size) and measurement stabilization. By use of microelectromechanical-system (MEMS) vibrating-beam technology, the authors introduce a borehole gravimeter that enables the recording of gravitational acceleration at very high sensitivity. The new resonant MEMS gravitational sensor is designed to sense mass in the subsurface such that time-lapse wireline-based surveys can be conducted to build a picture of fluid movements at relatively large distances from a wellbore, thereby enabling time-lapse or 4D gravity monitoring. The innovation of the three-axis gravimeter also allows the acquisition of directional information about the spatial movement of fluid, even when acquired from only a single borehole. The principles of measuring gravity in the borehole environment are provided in the complete paper. Libra Field. The field is one of the largest presalt oil discoveries in the Santos Basin. It was discovered in May 2010 with recoverable resources of approximately 9 billion bbl of oil. The first appraisal well was completed and tested in February 2015. At the time of writing, nine appraisal wells have been drilled. A high-density 3D seismic program was acquired over the Santos and Campos basins, and advanced seismic processing technology has also been applied. Reservoir Simulation. The geophysics for reservoir characterization and simulation of these carbonate reservoirs is still in its early stages, and little published knowledge about their properties exists. Nonetheless, since their discovery, 2D and 3D seismic technologies have been applied intensively for imaging the region structures and delineating the reservoirs.

Evaluation study on the effect of production and fluid reinjection towards reservoir using micro gravity in Kamojang geothermal field

Geothermal fluid exploitation is expected to cause physical as well as chemical changes to the reservoir of a geothermal field. This is what happened to Kamojang Geothermal Field which has been producing for four periods, starting from the initial production capacity of 30 MW (1983) which became 140 MW (1987), then 200 MW (2008) and 235 MW since 2015 up to now. To observe changes of subsurface condition, Microgravity, is called time-lapse gravity as well, as one of geophysical survey activity is carried out in order to obtain the reservoir changes in a wider view based on the changes of gravity value as a reflection of rock density changes due to the extracted fluid mass. Generally, the microgravity study result from 1984 until 2018 shows the existence of microgravity value changes which correlates to the amount of fluid mass produced and the water mass which was reinjected back into the reservoir. This can be observed from the Time-lapse Anomaly Map in accordance to each of the production period of Geothermal Power Plant (GPP) Unit 1, 2, 3, 4 and 5.

Time-Lapse Gravity Monitoring of Groundwater of the Sinai Peninsula

The Middle East is suffering from water scarcity in the arid/semiarid settings. The recent advance of technologies in the geophysical fields made groundwater monitoring possible from space. Time-variable gravity data and climatic model are utilized to monitor mass variations caused by groundwater changes over the Sinai Peninsula during the period 04/2002-07/2016.

Integrating time-lapse gravity, production, and geologic structure data in a gas reservoir study

Time-lapse gravity is most commonly used to monitor fluid movement and is especially useful when monitoring water encroachment in a gas reservoir. Although time-lapse gravity data are directly sensitive to density changes in reservoirs, it is still necessary to integrate multiple types of data with complementary information to enhance the time-lapse gravity interpretation. When monitoring water influx in a reservoir, the changes in water yield in production wells may directly indicate saturation changes with time and provide such complementary information about the areas of fluid movement. We have developed a workflow to invert a time-lapse gravity data set and production data to help monitor the edge-water encroachment through a case study at the Sebei gas field in western China. Three time-lapse gravity surveys were acquired between 2011 and 2013, resulting in three sets of time-lapse gravity difference data. Production data were also collected from 286 wells during the same period of time. We integrate the two data sets and the structural information in the reservoir through a framework of constrained time-lapse gravity inversion. In this workflow, we incorporate information from the production data into the inversion by converting the gas and water yield into a reference model. We also incorporate geologic structural information through spatially varying bound constraints. Through this approach, we construct a set of time-lapse density contrast models that are consistent with the time-lapse gravity data, production data, and structural information. The resultant density contrast models better delineate the regions of reservoir with increased water influx and also enable improved porosity estimation in the reservoir.

Annotation of Using Borehole Time-Lapse Gravity by Genetic Algorithm Inversion for Subsurface Modeling

We present the annotation to a genetic algorithm (GA) method for an inverse synthetic subsurface density model using surface and borehole time-lapse gravity data. The objective of the inversion is to find the boundaries of the object area and background, where one bit of the chromosome represents the densities. The model that was used in this paper was a simple homogeneous body anomaly and a simplified real water mass injection model in order to argue that the code is suitable for field modeling. We show the influences of the existence of borehole gravity data and location towards the inversion, where the result indicates that an additional good borehole location could increase the success rate up to 13.33% compared to without gravity borehole data for the simple model and up to 4.39% for the field model. The inversion produced the best results when the borehole positions were placed in a state of symmetry towards the body object's mass.

Time-lapse gravity monitoring of CO2 migration based on numerical modeling of a faulted storage complex

In this study, the performance of both surface and borehole time-lapse gravity monitoring to detect CO2 leakage from a carbon storage site is evaluated. Several hypothetical scenarios of CO2 migration in a leaky fault, and thief zones at different depths at the Kimberlina site (California, USA) constitute the basis of the approach. The CO2 displacement is simulated using the TOUGH2 simulator applied to a detailed geological model of the site. The gravity responses to these CO2 plumes are simulated using forward modeling with sensors at ground surface and in vertical boreholes. Results of inversion on one scenario are also presented. The surface-based gravity responses obtained for the different leakage scenarios demonstrate that leakage can be detected at the surface in all the scenarios but the time to detection is highly variable (10–40 years) and dependent on the detection threshold considered. Borehole measurements of the vertical component of gravity provide excellent constraints in depth when they are located in proximity of the density anomaly associated with the presence of CO2, thus discriminating multiple leaks in different thief zones. Joint inversion of surface and borehole data can bring valuable information of the occurrence of leakages and their importance by providing a reasonable estimate of mass of displaced fluids. This study demonstrates the importance of combining multiphase flow simulations with gravity modeling in order to define if and when gravity monitoring would be applicable at a given storage site.

Moving Average Filter untuk Memisahkan Efek Dangkal Anomali Gravitasi Time Lapse

[Title: Moving Average Filter for Time Lapse Gravity Anomaly Separation]. Interpretation of time-lapse gravity anomaly due to fluid injection in the reservoir is difficult when it mixed with shallow source anomalies. To solve the kind of problem, anomaly separation is performed with the method of Moving Average Filter. This study was conducted to obtain an effective method to separate the gravity anomaly of shallow sources on time-lapse gravity anomaly. Trials are conducted on anomaly models derived from reservoirs with three distinct depths that are 300, 600 and 900 meters. This forward model is then mixed with gravitational anomaly from the shallow source obtained from the field data. The mixed anomaly is then separated by a Moving Average filter. Results show that Moving Average filters can separate the shallow effect from the deep source anomaly and are effective up to a depth of 900 meters. The research is also beneficial for classroom learning in the computer programming class based on Matlab.

Hydrothermal fluid migration due to interaction with shallow magma: Insights from gravity changes before and after the 2015 eruption of Cotopaxi volcano, Ecuador

On August 14, 2015 Cotopaxi Volcano (Ecuador) erupted with several phreatomagmatic explosions after nearly 135 years of quiescence. Unrest began in April 2015 with an increase in the number of daily seismic events and inflation of the flanks of the volcano. Time-lapse gravity measurements started at Cotopaxi volcano in June 2015. Although minor gravity changes were detected prior to eruptive activity, the largest gravity variations at Cotopaxi were measured between October 2015 and March 2016, when other geophysical parameters had reached background levels. Inverse modelling of GPS data suggests a deep intrusion prior to the eruptive activity, while inverse modelling of post-eruptive gravity changes suggests variations in the volcano hydrothermal system. Deformation, seismicity, and gravity changes are consistent with the intrusion of a deep magmatic source between April and August 2015. Part of the magma rose from depth and interacted with the hydrothermal system, causing the phreatomagmatic activity and pushing hydrothermal fluids from a deep aquifer into a shallow perched aquifer.

Pemodelan Sesar Pasca Gempa Dengan Inversi Data Gravitasi Time Lapse

Time lapse or 4D gravity is the development of gravity methods with the fourth dimension is time. This method is characterized by repetitive measurements over a period of time using a high-precision gravity measuring device (gravimeter) which is supported by high accuracy measurement of position. In the case of a fault that triggers an earthquake, layers that have a different density between layers will have the effect of changing the measured value of the gravitational field on the surface. The difference in the value of measured gravity before and after the occurrence of a fault as a result of an earthquake is what gives the value of time lapse gravity. The model of position displacement of the layers before and after the earthquake, provides a model of density between times whose gravitational anomalies can be calculated by forward response. The response of time-lapse forward model (synthetic data) is then optimized to obtain an inversion model. Inversion programming is made using the Mathlab program which is based on the semi-infinite horizontal slab equation. The model parameters that are optimized are horizontal position, depth, and amplitude factor of faulted layers. The program testing on 3 different synthetic data shows that the fastest optimization was obtained in the 7th iteration step. The smallest RMS error from the 3 models was 1.23. 10-5. This inversion program can be applied to time lapse gravity anomalies results of measurements in the field in active fault zones before and after the occurrence of an earthquake

Time-lapse gravity and levelling surveys reveal mass loss and ongoing subsidence in the urban subrosion-prone area of Bad Frankenhausen, Germany

Abstract. We present results of sophisticated, high-precision time-lapse gravity monitoring that was conducted over 4 years in Bad Frankenhausen (Germany). To our knowledge, this is the first successful attempt to monitor subrosion-induced mass changes in urban areas with repeated gravimetry. The method provides an approach to estimate the mass of dissolved rocks in the subsurface. Subrosion, i.e. leaching and transfer of soluble rocks, occurs worldwide. Mainly in urban areas, any resulting ground subsidence can cause severe damage, especially if catastrophic events, i.e. collapse sinkholes, occur. Monitoring strategies typically make use of established geodetic methods, such as levelling, and therefore focus on the associated deformation processes. In this study, we combine levelling and highly precise time-lapse gravity observations. Our investigation area is the urban area of Bad Frankenhausen in central Germany, which is prone to subrosion, as many subsidence and sinkhole features on the surface reveal. The city and the surrounding areas are underlain by soluble Permian deposits, which are continuously dissolved by meteoric water and groundwater in a strongly fractured environment. Between 2014 and 2018, a total of 17 high-precision time-lapse gravimetry and 18 levelling campaigns were carried out in quarterly intervals within a local monitoring network. This network covers historical sinkhole areas but also areas that are considered to be stable. Our results reveal ongoing subsidence of up to 30.4 mm a−1 locally, with distinct spatiotemporal variations. Furthermore, we observe a significant time-variable gravity decrease on the order of 8 µGal over 4 years at several measurement points. In the processing workflow, after the application of all required corrections and least squares adjustment to our gravity observations, a significant effect of varying soil water content on the adjusted gravity differences was figured out. Therefore, we place special focus on the correlation of these observations and the correction of the adjusted gravity differences for soil water variations using the Global Land Data Assimilation System (GLDAS) Noah model to separate these effects from subrosion-induced gravity changes. Our investigations demonstrate the feasibility of high-precision time-lapse gravity monitoring in urban areas for sinkhole investigations. Although the observed rates of gravity decrease of 1–2 µGal a−1 are small, we suggest that it is significantly associated with subterranean mass loss due to subrosion processes. We discuss limitations and implications of our approach, as well as give a first quantitative estimation of mass transfer at different depths and for different densities of dissolved rocks.