Evaluation of the capacities of a field absolute quantum gravimeter (AQG#B01)

Quantum gravimeters are a promising new development allowing for continuous, high-frequency absolute gravity monitoring while remaining user-friendly and transportable. In this study, we present experiments carried out to assess the capacity of the AQG#B01 in view of future deployment as a field gravimeter for hydro-geophysical applications. The AQG#B01 is the field version follow-up of the AQG#A01 portable absolute quantum gravimeter developed by Muquans. We assess the instrument’s performance in terms of stability (absence of instrumental drift), sensitivity in relation to other gravimeters, and 5 hydrogeological mass changes. We discuss the observations concerning the accuracy of the AQG#B01 in comparison with a state-of-the-art absolute gravimeter (Micro-g-LaCoste, FG5#228). Repeatability is tested by instrument displacement between close-by measurement positions. We report the repeatability to be better than 50 nm.s−2. No significant instrumental drift was observed over several weeks of measurement. This study furthermore investigates whether changes of instrument tilt and external temperature and combination of both, which are likely to occur during field campaigns, influence the measurement of 10 gravitational attraction. We repeatedly tested external temperatures between 20 and 30 ◦C and did not find any significant effect. As an example of a geophysical signal, a 100 nm.s−2 gravity change is detected with the AQG#B01 after a rainfall event at the Larzac geodetic observatory (Southern France). The data agreed with the gravity changes measured with a superconducting relative gravimeter (GWR, iGrav#002) and the expected gravity change simulated as an infinite Bouguer slab approximation. We close with operational recommendations for potential users and discuss specific possible future field applications. While 15 not claiming completeness, we nevertheless present the first characterisation of a quantum gravimeter carried out by future users. Crucial criteria for the assessment of its suitability in field applications have been investigated and are complemented with a discussion of further necessary experiments.

(Micro-g-LaCoste, FG5#228) has been transported to and operated at the site. During the International Key Comparison of Absolute Gravimeters in 2017 (CCM.G-K2.2017), the FG5#228 showed a degree of equivalence of 3 nm.s −2 with the 12 other absolute gravimeters and 55 nm.s −2 uncertainty within 95% confidence (Wu et al. (2020)). In this study, the FG5#228 serves as a reference. In the Larzac observatory, the AQG#B01 and FG5#228 were operated on the same concrete pillar with approximately one meter distance between both instruments.

5
The AQG#B01 is the field version follow-on of the AQG#A01 described in Ménoret et al. (2018). It is based on the same measurement specifications and overall architecture but underwent a complete system redesign in order to meet outdoor operation requirements and to increase the stability of the measurement. The laser module and sensor head have been provided with an active thermal stabilization, allowing for a potential operation temperature range between 0 • C and 40 • C. Power consumption has been reduced to less than 500 Watt and battery operation is possible for field experiments. Improvement in ease of use 10 and transportability has been achieved with each element weighing 40 kg or less. The enclosure for the lasers and the sensor head has been made water-and airtight. Connectors and cables are suitable for field conditions and a reduction in the number of connectors further facilitates fast and efficient field set-up. As an example, shifting the AQG#B01 in the Larzac observatory from one pillar to another takes around 5 minutes for one person.
In late April 2020, additional AQG#B01 and FG5#228 measurements and tests were continued in the facilities of the labora-15 tory Géosciences Montpellier in the cellar, about 100 kilometers southwest of the Larzac site. The laboratory is in an urban area on the Montpellier University campus. Previous FG5#228 measurements show small gravity changes (less than 50 nm.s −2 over one year) (Jacob et al. (2008)) for a close-by site on campus. Environmental noise is monitored with a large band seismometer. Due to the Covid-19 lockdown, the environmental noise is largely reduced: less difference in noise level is seen between workdays, weekends, and public holidays. 20

Methods and experiments
The experiment timeline and data availability are displayed in Figure 1. The iGrav#002 data are available continuously. Software malfunctioning or updates and seismic events caused data gaps in the AQG#B01 series. In late January a seismic event made a restart necessary and caused no damage to the instrument. Improvements to avoid loss of measurements caused by these incidents are in progress. An instrument test was conducted remotely by the developer on March 23rd. Apart from those, 25 an offset of 100 nm.s −2 was observed to emerge in the AQG#B01 gravity time series before the second temperature test on 10/02/2020. The cause is still under investigation and the authors are in contact with the instrument developer. Additional monitoring variables registered during operation are being investigated. Up to this point, the main hypothesis is mechanical stress in the sensor head, acquired in-between temperature tests. data at dates marked in gray in later Janaury, early Feburary and late March were not considered in the analysis, due to inconsistencies during the experiments, as explained in section 4.4.2. The thin blue line represents the residuals from the mean observatory temperature to illustrate the temperature tests. One test has been carried out at reduced temperatures and four at increased temperatures. Grey diamonds during the AQG-B01 series mark the dates of tilt tests carried out on the AQG-B01.

Drift, accuracy and sensitivity
The sensitivity is assessed by calculating the Allan deviation (Allan (1966)) as a measure of frequency stability over the gravity data series. The Allan deviation is calculated for the three gravimeters (FG5#228, AQG#B01, and iGrav#002) after classical post-processing: calibration and drift correction of the iGrav#002 data set, correction for solid Earth and ocean loading tides using Tsoft (Camp and Vauterin (2005)), atmospheric pressure and polar motion for all data sets. Polar coordinates were 5 obtained from the International Earth rotation and Reference systems Service. Site-specific combined ocean and solid Earth tidal parameters had been estimated with ETERNA (Wenzel (1996)) software based on long-term iGrav#002 time series (Fores et al. (2016a)). Gravity residuals refer hereafter to the processed gravity data set.
The influence of global, non-local hydrological and atmospheric gravity effects on the Larzac site was estimated using the EOST loading service (Boy and Hinderer (2006); Boy and Lyard (2008); Boy et al. (2009)) applying the model GLDAS/Noah 10 (v2.1) (Rodell et al. (2004)). Gravity residuals obtained from AQG#B01 and iGrav#002 were related to local cumulative precipitation obtained from on-site rain gauges to assess the detectability of small hydrogeological mass changes. A 1D hydrological model using rainfall as input describes the gravity changes due to hydrological mass changes adequately (Fores et al. (2016a)), 5 https://doi.org/10.5194/gi-2020-22 Preprint. Discussion started: 21 August 2020 c Author(s) 2020. CC BY 4.0 License. hence to display the gravity changes caused by rainfall an infinite homogeneous Bouguer anomaly was assumed. The Bouguer plate was calculated according to: (1) G refers to the gravitational constant, ρ to the density of water and the plate thickness H refers to the cumulative rainfall.
The Bouguer plate equivalent was corrected for the estimated averaged daily deep percolation discharge of one mm per day 5 (Fores et al. (2018)). The investigated precipitation event took place during the winter months, evapotranspiration was thus not considered. The FG5#228 measurements provide absolute reference points to assess any drift with time in the AQG#B01 time series. The period between the 28/11/2020 and 25/01/2020 was used for drift assessment, as numerous tests (tilt, temperature) were conducted afterwards. Daily averaged residuals are compared in order to assess the accuracy of the AQG#B01.
The difference of effective measurement height requires a correction for the vertical gravity gradient when the accuracy of 10 the AQG#B01 is estimated. For this set-up, the AQG#B01 effective measurement height was at 65.1 cm, the FG5#228 at 121.77 cm. The horizontal gravity differences between the FG5#228 and AQG#B01 measurement locations in Larzac and Montpellier have been estimated with a relative Scintrex CG5 and CG6 gravimeter, at 1.2 m height difference. Estimated vertical gravity gradients at the Montpellier site were found to be approximately -2.9 kE (1 kE = 10 nm.s −2 cm −1 ). In the Larzac observatory estimated vertical gravity gradients on pillar 1 (FG5) and 2 (AQG) are -3.225 and -3.220 kE, respectively, averaged over one 15 year of monthly measurements (Cooke et al., in preparation). AQG#B01 and FG5#228 gravity residuals were thus transferred to the same height by correcting for a vertical gravity gradient of -3.22 and -3.225 kE in the Larzac Observatory and for -2.86 and -2.89 kE in the laboratory in Montpellier, respectively.

Adjustment of ambient temperature in the observatory
The observatory is kept at relatively stable 24 • C in the weeks before and in between the experiments. The AQG#B01 was 20 operated during five periods of modified ambient temperature with periods of standard temperature in between ( Figure 1). The temperature in the observatory was changed by adjusting the air conditioning device. The first temperature test comprised a reduced temperature, followed by four tests of higher temperatures, relative to the 24 • C default temperature listed in Table 1.
During the first period of increased temperature, an elevated noise level and interruption due to the seismic event were observed by both the iGrav#002 and the AQG#B01 and these periods were therefore not considered in the analysis.  strument's tilts in x and y are adjusted manually to reach a certain angle θ and the raw gravity data are then adjusted along a function of 1 cos(θ−θ0) , in order to evaluate θ 0 , corresponding to the real vertical axis. The offset coefficient has been tested in the Muquans facilities in Talence (France) and has since been redone twice at the Larzac observatory. Furthermore, it was investigated whether the obtained offset coefficient had changed over time or had shown any response to temperature changes.
The offset calibration test on 19/02/2020 was carried out in the GEK at an increased temperature of 28 • C and 05/03/2020 at 5 22 • C.

Manual tilt deregulation under adjusted ambient temperatures
Temperature changes or temperature gradients may influence mechanical parts and tilt the instrument. To investigate possible interactions between temperature and tilt and to ensure reliable application of their corrections, manual tilt deregulation was carried out during phases of modified ambient temperature. Between 09/12/2019 and 13/12/2019, manual tilt deregulation in 10 x and y during room temperature was tested. On March 11th the room temperature was modified from 22 to 30 • C and on the same day, the tilt in x-direction was manually set to 0.5 mrad.

Repeatability
On April 17th, 2020 the AQG#B01 was transported to the facilities of the laboratory Géosciences Montpellier on University campus and operated in the basement of the building. The distance between the Larzac observatory and Montpellier is about 15 100 km and there is 640 m difference in altitude. The transport to Montpellier was the second displacement of the AQG#B01 after its first delivery to the Larzac observatory in November 2019. This implies the turn-off, deconnection, displacement and cold restart of the instrument. The data were compared to FG5#228 measurements at both sites. Small-scale repeatability was assessed using repeated gravity measurements on the same position in the gravity lab in the basement of Géosciences Montpellier. Vertical gravity gradients were additionally estimated with a relative gravimeter (Scintrex CG6).

Coriolis effect
Gravimeters are sensitive to a Coriolis shift, the Sagnac effect caused by the Earth's rotation. This effect can generate an additional bias in quantum interferometers. The horizontal atomic velocity component generates an additional Coriolis acceleration depending on the E-W direction. This leads to a possible gravity bias (Peters et al. (2001), Louchet-Chauvet et al. (2011)). By symmetrical construction (hollow pyramidal reflector and location of the detection photo-diodes), horizontal atomic velocities 5 are reduced and the AQG#B01 should not be sensitive to the Coriolis effect. We assess the potential residual Coriolis effect in the AQG#B01. By rotating the device by 180 • measurements, two opposite orientations are obtained, hence a change in sign of the Coriolis acceleration is expected. As in Louchet-Chauvet et al. (2011), the tests were carried out under the assumption that parameters do not change between the measurements. The same set-up was thus kept constant to rule out other systematic effects. Coriolis AQG#B01 measurements last at least 24 hours to reduce the effect of residuals after tidal correction. The sensitivity of the AQG#B01 is firstly evaluated by statistical time series analysis in comparison with other gravimeters and secondly by direct monitoring of natural gravity changes. It was calculated for the first week of December 2019 using 10 minute AQG#B01 data and one-minute iGrav#002 data ( Figure 2). The first week in December 2019 shows no rainfall and no 15 instrument tests were performed. At an integration interval of one hour, the sensitivity of the AQG#B01 reached 10 nm.s −2 , the iGrav#002 shows a higher sensitivity at short time scale, but an increase at long time due to environmental noise and tides residuals. At 24h the AQG#B01 approaches the sensitivity of the iGrav, and averages around 5 nm.s −2 for this time span. (grey), 10-minute data for the AQB-01 (red). The irregular sampling frequency of the FG5-228 (100 drops every 10 seconds, then 2600 s break) was taken into account by plotting the short-term Allan deviation (< 1000s for 10 s data) and the long-term (1 h data) separately (both blue). The horizontal blue dashed line shows the sensitivity benchmark of 10 nm.s −2 , the dark green vertical dashed line signifies the integration period of 1 h, the light green one that of 24 h.
The analysis of the Allan deviation showed that for averaging periods of a few hours the iGrav#002 is the more sensitive instrument for this data set. The AQG#B01 and the iGrav#002 converge to show similar sensitivity of clearly better than 10 nm.s −2 at a 24 h averaging period. For shorter integration intervals the sensitivities of the FG5#228 and AQG#B01 are comparable and lower than that of the iGrav. The AQG#B01 does not achieve the sensitivity of laboratory quantum gravimeters that have achieved 2 nm.s −2 in less than 2000 s (CAG; Gillot et al. (2014)) or a mobile quantum gravimeter for which 0.5 5 nm.s −2 after one day have been reported (GAIN; Freier et al. (2016)).
For measurements longer than one day, the AQG#B01 is likely to be equally sensitive as the iGrav#002. To obtain values closer to the possible highest sensitivity, a prolonged measurement of several weeks during a low noise period of stable weather conditions and little human interventions is required, as it would be possible for example during summer months in the Larzac observatory. The Allan deviation of the AQG#B01 data recorded in Montpellier showed a sensitivity of approximately 20 10 nm.s −2 after one h, after 24h it was below 10 nm.s −2 . This decrease in sensitivity for the urban site compared to the Larzac observatory can be explained by the higher level of environmental noise in the university building.
A sensitivity of 10 nm.s −2 is achieved in one hour in a naturally low noise environment and a sensitivity of 20 nm.s −2 in an urban environment. In the context of characterising the AQG#B's sensitivity, the use of rubber pads below the tripod feet to reduce the effect of ground vibrations was assessed for future field experiments. Figure 3 shows that the Allan deviation is reduced for measurements of less than an hour. At one hour duration, there is no significant difference, for longer integration times there is likely no major improvement. This needs to be reassessed for longer series and at periods of higher environmental noise as the activity at the university was reduced during the COVID19-lockdown. Figure 3. Comparison of sensitivity with (orange) and without (grey) placing the AQG-B01 tripod on rubber pads. The measurement without rubber pads were conducted on 23-26/04/2020, without pads on 28/04/2020. 4.2 Sensitivity to hydrogeological gravity changes 5 A significant increase in g of ∼ 80 nm.s −2 between the 13/12/2020 and the 27/12/2020 has been detected by the three gravimeters. As can be seen on Figure 4 (panel (a)) daily gravity residuals of all of them show high resemblance in their temporal variations and differ within their error margins. Figure 4 panel (b) shows the series of rainfall events in 2019. All three gravimeter signals follow the increase in gravity caused by the rise in soil water content in the aftermath of the rainfall events. The gravity time series continues to increase even 10 after rainfall stops. This is expected since the infiltrating water moves further into the gravimeter's spatial sensitivity which can be described as two flat cones above and below with the instrument in the centre. Umbrella effect related differences between AQG#B01 and iGrav#002 could hence be up to several tens of nm.s −2 . As the umbrella effect depends on the initial conditions and the previous rainfall events, it is difficult to determine the sign of the relative offset between the AQG#B01, FG5#228 and iGrav#002 series without further information.
To summarise, the AQG#B01 clearly measures the gravity increase of less than 100 nm.s −2 after the rainfall event in the same range as the iGrav. The gravity changes are coherent with previous studies at the site. The differences between the iGrav#002 and AQG#B01 data can be explained by limits of the sensitivity of the AQG#B01 and the heterogeneity of the hydrogeological context in a Karst area.

Accuracy, repeatability, and drift
The differences between the AQG#B01 and the FG5#228 during 2 months yielded an estimated, statistically insignificant drift of -0.02 ± 0.04 nm.s −2 per day. A longer measurement is currently in progress to investigate a potential long-term drift. Establishing the complete accuracy budget is a complex task for a new instrument and remains work in progress. A first approximate estimation of the accuracy is done by comparing the AQG measurements with the ballistic gravimeter (FG5#228) in the Larzac 5 observatory. Daily AQG#B01 and FG5#228 gravity residuals show high resemblance in their temporal variations and differ within their error margins (Figure 4). The difference between AQG#B01 and FG5#228 based on 13 measurements (daily averages) between December 2019 and April 2020 in the Larzac observatory is on average 110 nm.s −2 with a standard deviation of 31 nm.s −2 , with the FG5#228 values being smaller than those measured with the AQG#B01. For the Montpellier laboratory, the difference between both instruments is based on 24h averages of ten AQG#B01 measurements between 27/04/2020 and 10 14/05/2020 and one FG5#228 measurement on 10/06/2020. The difference showed 44 nm.s −2 with a standard deviation of 66 nm.s −2 , with the FG5#228 values being higher than those of the AQG#B01. Absolute comparison between both instruments was not directly possible due to the set-up on different pillars and is impacted by the uncertainty related to the vertical gravity gradient (VGG) correction. An offset in vertical gravity gradient correction between FG5#228 and AQG#B01 of 10 E yields for the difference in height between the two instruments' sensors (δ height = 56.67 cm) 5.7 nm.s −2 of uncertainty. The VGG for the pillars in the observatory can be estimated by repeated relative gravimeter measurements on different heights and their uncertainty has been estimated to be around 20 E (Cooke et al., in 5 preparation). Hence, about 11 nm.s −2 of uncertainty is due to the fact that the VGG cannot be estimated more precisely up to this point. The possibility of a more precise estimation of the VGG based AQG measurements on tripods is being discussed and in preparation.
Small-scale repeatability tests were only carried out in Montpellier. Table 2 shows an average small-scale repeatability of 3 nm.s −2 with a standard deviation of 25 nm.s −2 for repeated measurements on the same point and orientation after returning 10 from displacements and other experiments.
AQG#B01 was operated on two measurement points within the same room in the Géosciences laboratory at about one meter distance, of which one serves FG5#228 measurements. At this short distance, no considerable horizontal difference in g is expected. CG6 relative gravity measurements were carried out with two instruments (CG6#120 and CG6#125) on 26/03/2020 on both points and found a negligible difference in g of 2 ± 6 nm.s −2 . The difference in g measured with the AQG#B01 15 between point 1 and point 2 is 15 ± 48 nm.s −2 . The measurement on point 2 was carried out using rubber pads under the tripod which added a height of 1.2 cm. This height difference was corrected with 35 nm.s −2 based on an estimated vertical gravity gradient obtained from CG6 measurements.
To summarise, these first results show a repeatability better than 50 nm.s −2 and no detectable drift over 2 months operation.
No impact of transport and displacement such as mechanical relaxation on the gravimeter was observed for small (one m) or temperatures. The data showed that the tilt calibration is likely to be independent of temperature and to stay stable over time.

Influence of temperature
During the temperature experiment AQG#B01 residuals did not show any statistically significant correlation with external temperature. Gravity residuals, external temperature and tilts are displayed in Figure 5. The AQG#B01 showed no significant During the second temperature test an insulating cover around the sensor head had not been removed resulting in an increase of temperature in the sensor head above nominal operation conditions. AQG#B01 and iGrav#002 residuals differ less than 50 nm.s −2 . The reduced temperature test (20 • C; January 20-28th) did 25 not yield any significant gravity response to increased external temperatures. In March and April, the tilts stay stable before, during, and after the temperature test and no remarkable shift between the residuals of g obtained with the AQG#B01 compared to those obtained with the iGrav#002 is observed ( Figure 5).

Combined tilt and temperature tests
Manually deregulated tilts during temperature changes did not show a visible change in g recorded by the AQG#B01. As can be seen in Figure 5, the tilts showed a minor response during the last temperature test in early April. No impact on the corrected g value (with a tilt correction of ∼ 1000 nm.s −2 ) is observed. Tilts return to values close to zero after a change in temperature.
Simultaneous manual tilt deregulation and room temperature change did not lead to any clear shift of the difference in g 5 between the value of the AQG and that of the iGrav. This result suggests that the measurement of g is not impacted by temperature. It cannot be ruled out that tilts of more than one mrad require a different correction than small tilts. It is thus recommended to keep the sensor head well levelled during operation. The gravity series obtained with the AQG#B01 before, during, and after an elevated temperature of 30 • C in March and April 2020 show no impact of these. To reproduce these findings, further temperature experiments and larger ranges should be carried out, potentially exploring also much lower temperatures. To assess the potential impact of the Coriolis effect caused by changes in the sensor head's orientation we carried out measurements rotated by 180 • compared to the default position. As can be seen in Figure 6, the averaged gravity residuals for the two orientations show a δg of approximately 150 nm.s −2 relative to the FG5#228 measurement. These results are much higher than expected, as it has not been observed in other AQG devices (Muquans, 2020, personal communication). These values are 5 higher than the estimation of the Coriolis effect for the CAG (LNE-SYRTE), which yielded 4 nm.s −2 uncertainty with peaks of up to 60 nm.s −2 as a combination of several uncorrected effects (Louchet-Chauvet et al. (2011)).
The sign of the Coriolis offset on the AQG#B01 as compared to the absolute gravity measurement by the FG5#228 has an important implication for the accuracy assessment and the interpretation of the differences between the two instruments: The FG5 measurement lies between the AQG#B01 values for the two orientations during the tests in Montpellier. FG5#228 10 measurements were higher than AQG#B01 measurements in the Larzac observatory. The average difference between both instruments hence likely requires to be increased by ∼ 75 nm.s −2 , depending on the orientation of the AQG#B01 sensor head.
Repeated and additional orientations (90 • and 270 • ) are work in progress, as well as in comparison with other gravimeters.
The authors are in contact with the developer for in-depth instrument tests. Up to this date, the exact source of this change in g has not been identified yet. It is thus recommended to pay attention to sensor head orientation during operation. It will be necessary in the future to evaluate through repeated tests if the impact of Coriolis is stable over time and and whether a Coriolis correction can be established according to the orientation. The Coriolis effect needs to be included in the complete uncertainty 5 budget.

Conclusion and perspectives
In this study, we show the results of instrumental tests aimed at the characterization of the AQG#B01 for field applications.
The AQG#B01 has proven itself as a reliable instrument in controlled laboratory conditions. Over two months no significant drift was observed and temporal variations are in coherence with the MGL-FG5. Its sensitivity after 24 h of data integration is 10 close to that of the iGrav. In low noise environments, the AQG#B01 showed a sensitivity of 10 nm.s −2 after 1 h. AQG#B01 g residuals showed no correlation with manually increased tilts nor increased temperatures. An offset compared to the other gravimeters occurred and its causes are under investigation, as is the accuracy of the AQG#B01.
The obtained results further suggest its suitability for field studies, upon further testing and validation. It is suitable for operation at least within a temperature range between 20 and 30 • C over several days and weeks. Larger temperature ranges 15 are possible but have not been systematically tested yet. Tilt correction is likely to be applied correctly even for relatively large tilts and during periods of higher temperatures.
The AQG#B01 detected gravity changes caused by hydrology in the same order of magnitude as the iGrav. The resemblance between AQG#B01 and iGrav#002 residuals concerning their response to a rainfall event demonstrates the AQG#B's capability to detect small transient mass changes. This speaks for the applications of the AQG#B01 in hydrogeophysical studies among showed that relative gravimetry with a CG5 is highly sensitive to temperature. Correcting for temperature induced effects shed new light on apparent spatial gravity differences measured in the field. It is possible that the spatial water storage heterogeneity suggested by Jacob et al. (2010) needs to be re-interpreted in view of remaining, uncorrected effects.
The applications in the mentioned studies show the potential gain in precision and time saved provided by the AQG. The AQG#B01 allows to combine two instruments in one. In absence of a detectable drift, regular calibration is not required. In principle, no repeated loops would be necessary, as no gravity ties need to be established. The need for another indoor reference gravimeter becomes obsolete. High precision gravity acquisition is possible with this new movable instrument. It is easy to set-up and use without the need for operation and maintenance by an expert, as for the FG5. The survey time investment and 5 data treatment could be hence reduced remarkably for spatial gravity mapping. The results so far suggest a sensitivity between 10 and 20 nm.s −2 after 1 hour. These first results are promising that the AQG#B01 could reach significantly higher precision than relative gravimeters while being transportable. Even if the sensitivity of the AQG#B01 during outdoor operation still needs to be investigated, the results suggest reliable operation in different temperatures, very likely reaching a higher sensitivity than that of relative gravimeters after only one h of measurement. The repeatability has been quantified as better than 50 nm.s −2 .

10
Our study also revealed important precautions that need to be taken. The first results on the Coriolis effect suggest that for repeated studies the same orientation of the sensor head needs to be kept.
Its time efficient deployment offers new possibilities for natural hazard monitoring and potential early warning systems, some of which are already under investigation with the AQG#A. Joint absolute and relative gravimetry monitoring of volcanic activity are studied at Mt Etna (Carbone et al. (2020)). Another recent project focuses on the AQG on a mobile facility for a 15 hydrological extreme event task force (Reich and Güntner (2020)). The observatory tests under controlled conditions aimed at singling out the effects of ambient conditions, mainly temperature. The next step is clearly to carry out tests outside the building. Repeated displacements between other observatory locations should be carried out to quantify the repeatability of the measurement. The estimation of the vertical gravity gradients by operating the AQG#B01 on two different heights would add another application to the instrument's repertoire.