Recent Activity of Nyiragongo (Democratic Republic of Congo): New Insights From Field Observations and Numerical Modeling
Abstract
Nyiragongo volcano is known for its active lava lake and for socioeconomic issues arising from future possible eruptive events having major impacts on the community living in the Virunga region. The 2020 field expedition inside the summit crater has allowed the collection of unprecedented field observations to state on the current activity. Since the February 2016 intracrater event, the crater floor level has been rising much faster than during the 2010–2016 period. The current activity is reminiscent of the 1970–1972 and 1994–1995 periods preceding the lava lake drainage events in 1977 and 2002. Numerical simulations, successfully validated with data over the past 30 years, show that the rising of the crater floor could slow down in the next months/years and reach a critical equilibrium. Based on the past eruptive history and on the current activity, a flank eruption in the March 2024 to November 2027 interval could be a possible scenario.
Key Points
- Current eruptive activity at Nyiragongo mirrors that observed in 1970–1972 and 1994–1995
- Activity forecasting is backed up by simulations of a mathematical model
- Based on the forecast failure model, a new lava lake drainage in the March 2024 to November 2027 interval could be a possible scenario
Plain Language Summary
Nyiragongo volcano hosts within its crater the world's largest continuously active lava lake. This constitutes major socioeconomic issues which arise from future possible eruptive events having a catastrophic impact to the Virunga volcanic region's growing community within the Democratic Republic of Congo (DRC) and Rwanda, particularly the cities of Goma and Gisenyi with about 1.5 million people living at its foot. The 2020 field expedition inside the summit crater allowed the collection of unprecedented field observations to state on the current eruptive activity. Since a new and unusual eruptive event occurred on 29 February 2016, the crater floor level has been rising much faster than during the 2010–2016 period. The current activity is reminiscent of the 1970–1972 and 1994–1995 periods preceding the lava lake drainage events in 1977 and 2002, respectively. Numerical simulations, successfully validated with data over the past 30 years, show that the rising of the crater floor could slow down in the next months/years and reach a precarious balance. Based on the past eruptive history and on the current activity, we stress that a flank eruption in the March 2024 to November 2027 interval could be a possible scenario.
1 Introduction
Increasing understanding of magma transfer from Earth mantle to surface is a challenging goal involving the need of a correct understanding of the evolution in time and space of eruptive activities (Boudoire et al., 2019; Poland et al., 2012). It is even more fundamental in (very) low- to middle-income countries where volcanic lands tend to favor socioeconomic development (Shoji & Takahashi, 2002). Unfortunately, environmental, economic, and/or political conditions are often a hindrance to the development of adequate monitoring strategies. The Virunga volcanic province, on the Eastern African Rift (EARS), is a perfect example of such a place (Figure 1a). The site needs a greater understanding and monitoring of the volcanic activity, which is as important as the challenge to face major implications arising from the densely populated areas in the vicinity of the volcano (Balagizi et al., 2018; Cuoco, Spagnuolo, et al., 2013; Cuoco, Tedesco, et al., 2013; Spampinato et al., 2013; Tedesco, 2002/2003; Tedesco et al., 2010). Among the eight volcanoes hosted in the province, Mount Nyiragongo hosts within its crater the world's largest continuously active lava lake (Durieux, 2002/2003). In the last five decades, Nyiragongo's lava lake was drained catastrophically twice, in 1977 and 2002 (Tazieff, 1984; Tedesco et al., 2007) from flank eruptions. These eruptions have raised important civil protection issues (400–1,500 fatalities in 1977; 250 in 2002; Tedesco et al., 2007). During the last event, the lava flowed within the city of Goma (Figure 1a) and caused major socioeconomic disruption by leaving 120,000 people homeless (Baxter et al., 2002/2003; Tedesco et al., 2007) and destroying more than 80% of the economic infrastructure.
The sequential increase in lava lake volume and rapid fissure eruptions are likely responsible for the peculiar morphology of the crater (Durieux, 2002/2003; Tazieff, 1984). Its current morphology is set by three distinct volcanic platforms: P1 (1977 crater floor), P2 (1995–2002), and P3 (2002 to present) (Figure 1). On 29 February 2016, the relative equilibrium between magma feeding and convection of the lava lake, which has characterized the volcanic activity after the 2002 eruption (Figure 1b), was disrupted by the opening of an intracrater fracture on the most eastern side of the crater (Balagizi et al., 2016; Valade et al., 2018). The fracture favored the appearance and growth of a spatter cone. On February 2017, the spatter cone was 30 m high and 70 m large at its base (Figure 1c). Its activity contributed to raise the level of P3 by successive lava flows invading the platform and modifying the dynamics of the whole volcanic system by pouring several tens of millions of cubic meters of lava into the lava lake (Burgi et al., 2018). In February 2020, 3 years later, the P3 level has risen by more than 100 m (Figure 1d), about three times faster than during the period 2010–2016. Such a fast increase is a putative indication of a change in the regime, which dominated the 2002–2016 eruptive phase. The paper aims to use the field observations of the crater floor level together with numerical model to decipher magma processes and suggest (new) monitoring strategies, in particular in the place where deploying and maintaining instrumental networks is highly challenging.
2 Field Observations and Measurements
We performed a field expedition inside the summit crater from 10 to 19 February 2020. During this field trip, a comprehensive series of field observations (Text S1) were fulfilled to document the physical configuration of the summit active crater and the rate of magmatic activity within the crater. Attention was focused on four peculiar features of the eruptive activity: the spatter cone, the lava fall (the cascade of lava) fed from the spatter cone and pouring into the lava lake, the lava lake itself, and two active lava ponds flowing in the northern part of the crater (Figure 1e).
2.1 Spatter Cone
The spatter cone is 45–50 m high and about 100 m large at its base. It consists of a tilted “initial” part (~32°) for about half of the height and becomes subvertical near the top (Figure 1d). On the northern side of the spatter cone, an elongated mound is piling up from spattering. At the time of observations, this mound culminated at a height of 17 m and gently sloping over 80 m (Figure S1). A continuous stirring of lava was visible in the U-shaped open vent (Figure 1e). Lava flows continuously expanded out at the base of the cone and poured into the lava lake through an underground tunnel. Intermittent effusive activity from the summit of the cone since its formation in 2016 (Valade et al., 2018) resulted in piling up lavas around and filling P3 to the point of reaching P2 and overflowing an area of about 16,000 m2 (Figure 1d). This new morphology makes it possible to access P3 and thus allows an easy access to the lava lake banks.
2.2 Lava Fall
A lava tunnel of about 180-m length with a slope of 7° connects the active spatter cone to the eastern rim of the lava lake. At the end of this tunnel, the lava cascades from a height of 1–2 m, depending on the lake level, into the lava lake. The outflow is difficult to precisely estimate due to the intense degassing and lack of visibility. However, based on the rare sequences of images of the lava fall captured with a standard camera (Video S1), we can estimate a maximum surface velocity of 3 m/s. Assuming a semicircular channel of radius r = 1 m (visual estimation), and the ground slope angle α = 7°, we apply a modified Jeffreys' equation to estimate a volumetric lava flux of 4 m3/s (Text S2; Jeffreys, 1925; Lev & James, 2014). This value is in the range of volumetric flow rates calculated for pahoehoe lava (Gregg, 2017; Rowland & Walker, 1990) and fits also with those reported for lava tubes of similar geometry (Melnik, 2017).
2.3 Lava Lake
The lava lake preserves its slightly elliptical shape with the long E–W axis measuring approximately 230 m, and 210 m for the N–S axis, for a total surface of approximately 40,000 m2. These values are about 10% smaller than those reported by Burgi et al. (2018) for September 2017 and by Barrière et al. (2019) for March 2018. Similarly, these values do not follow the linear progression that has taken place since the reappearance of the lava lake in 2002 (Burgi et al., 2014). Moreover, the position of the lava lake eastern rim has shifted toward the south-west by about 50 m between 2007 and 2015 and about 20 m toward the north-east between 2015 and 2020 (Figure 2). Since September 2017, concomitantly with the lava lake level, P3 has risen by more than 100 m. In February 2020, the level difference between P2 and P3 was in the range 0–25 m, with an overlapping of P2 and P3 around the spatter cone and the maximum measured in the western part of the crater. This fast evolution of the crater floor elevation is not unusual at Nyiragongo (Burgi et al., 2014). It is striking to note that a similar behavior was observed in the years preceding the complete drainage of the lava lake in both 1977 and 2002 events, with three phases (φ1, φ2, and φ3 on Figure 2e). A first phase (φ1), with a certain stabilization of the crater floor level between −400 and −300 m (relative to the crater rim), was observed in 1962–1970, 1985–1994, and 2010–2016. A second phase (φ2) of fast increase of the crater floor level was observed in 1970–1972, 1994–1995, and since 2016. Interestingly, these increases marked a change in the volcanic activity of Nyiragongo. On March 1971, the presence of vents emitting thick vapor and gaseous plumes was reported (Denaeyer, 1973; Durieux, 2002/2003), when no significant activity was observed before. In 1994 and 2016, intracrater eruptive activity started (Burgi et al., 2018; Tedesco, 2002/2003). A final phase (φ3) with the crater floor elevation reaching a plateau between −150 and −250 m (relative to the crater rim) was observed, occurring in the years preceding its complete drainage by flank eruption and lasting respectively from 1972 to 1977 and 1995 to 2002.
2.4 Lava Rises and Pits
In the NW part of P3, about 500 m away from the spatter cone and 25 m beneath its basis, several tunneled active lava flows form lava rises and small ponds (Figure S2), which have the characteristics of lava-rise pits (Walker, 1991). These ponds are fed by tunnels starting from the spatter cone as testified by the presence of many skylights. The largest pond, oriented NE–SW, is 80 m long and about 25–30 m wide and located 450 m to the west of the lava lake. The smallest one, oriented E–W, is 50 m in length and 20 m wide and located 200 m north of the lake. Driven by slow and nonconvective movements, lava flows from these two ponds into the lava lake. The entry points within the lava lake are observable when the level of the lake slightly decreases after degassing phases. The flux of these two inflows is estimated at 0.015 m3/s on average (Text S2). They represent only a secondary feeding compared to the lava fall in the eastern part, which is directly connected to the spatter cone.
3 Modeling and Discussion
The intracrater fracture of 29 February 2016 marked an important break in the eruptive activity of Nyiragongo volcano (Burgi et al., 2018). Coupling field observations with numerical modeling could shed new light on the evolution of the magmatic activity and its impacts on the morphology of the summit crater.
3.1 Magma Flux Modeling
By using the equations set for magma propagation through dikes (Gonnermann & Taisne, 2015; Gudmundsson, 2016), the evolution of the volumetric magma flux feeding the spatter cone since the opening of the intracrater fracture (possibly due to higher input of fresh magma which triggered a dike propagation) on 29 February 2016 may also be simulated (Figure 3b). The determinant factors in these calculations are the overpressure in the magmatic reservoir feeding the dike and the length of the dike (Traversa et al., 2010). In our case, the volumetric magma flux results from a magmastatic equilibrium between (i) the pressure of the shallow reservoir that is dependent on the accumulation of sinking degassed magma that increases the buoyancy force (Equation 2) and (ii) the spatter cone elevation that represents the top of the dike and which increases with P3 level. Consequently, in our model, both the overpressure and length of the dike increase in time. Our simulation shows that the volumetric magma flux rate feeding the spatter cone is expected to vary between 2 and 6 m3/s in time (Figure 3b). It is fully consistent with our estimation of 4 m3/s made during the February 2020 field expedition. However, this flux might have been much higher in an early phase of the spatter cone (Valade et al., 2018), and the curve in Figure 3b represents a leveling of the peak activity values.
3.2 Implications for the Morphology of the Summit Crater
Between 2017 and 2020, the crater floor quickly filled as lava flows piled up to almost 100 m high (φ2 on Figure 2e). This is three times more than what was observed between 2010 and 2016 (φ1 on Figure 2e). Knowing that the average diameter of P3 is 800 m (and did not change significantly between 2017 and 2020), the increase of the level of P3, including the lava lake, represents a lava input of about 50 Mm3, which in turn corresponds to an average excess volumetric magma flux of 0.6 m3/s in the 2017–2020 period. Together with the 0.6–3.5 m3/s magma flux necessary to keep the lava lake in melted state (Burgi et al., 2014), we estimate that the current volumetric magma flux at Nyiragongo falls in the range 1.2–4.1 m3/s, that is, 0.04–0.13 km3/yr. This volumetric magma flux is comparable with values reported for other lava lakes as at Kilauea (0.02–0.28 km3/yr; Anderson & Poland, 2016; Dvorak & Dzurisin, 1993) and Masaya (0.13–0.25 km3/yr; Zurek et al., 2019).
In this study, we estimated a maximal volumetric magma flux of 4 m3/s from the spatter cone during our stay. This magma flux feeding the spatter cone is expected to impact the dynamics of the lava lake by decreasing the upward feeding rate (0.6–3.5 m3/s) necessary to keep it in a molten state (both the lava lake and the dike being fed by the same shallow magma reservoir; Burgi et al., 2018). A natural consequence would be a decrease of the size of the lava lake to compensate the heat lost from the surface by radiation and convection. This effect is observed between 2015 and 2020 (Figure 2a). The eruptive activity of the spatter cone could also have an impact on the position of the lava lake. The surface motion of the lava lake is thought dominantly radial with a hot western zone (upwelling region) and a cooler eastern area (Lev et al., 2019; Spampinato et al., 2013) that may have triggered the westward shift observed between 2007 and 2015 (Figure 2a). However, it is possible that the development of the spatter cone in the eastern part of the crater in February 2016 has contributed to inverting the displacement of the lava lake between 2015 and 2020 (Figure 2a).
If the volumetric magma flux does not change, our numerical simulation predicts that the lava will overflow entirely P2 by February 2021. In the longer term, if no tectonic and/or magmatic event breaks the current magmastatic equilibrium, our simulation anticipates that (i) the eruptive activity of the spatter cone should progressively decrease until its total inactivity by the first quarter of 2022 (Figure 3b) and (ii) the overpressure in the shallow magma chamber would plateau at 19 MPa. This value is in the range of those (15–20 MPa) reported at Kilauea during the 2018 rift zone eruption (Anderson et al., 2019). According to our model, by 2025, the spatter cone will have discharged 600 Mm3 of lava in the lava lake and about 20 Mm3 on P3, contributing to 50 m on the P3 level increase since February 2016 (in complement to the lava lake overflows). In turn, the elevation of the lava lake should stabilize (and even slightly decrease) between −250 and −260 m (relative to the crater rim) in the next few years (Figure 3a). A similar scenario was observed in 1972–1977 and 1995–2002 (φ3 on Figure 2e), that is, in the years following a fast increase of the elevation of the lava lake and preceding its complete drainage. By the first quarter of 2025, the lava lake should contain about 16 Mm3 for a total of 2 km3 of intruded material beneath the edifice that has been accumulating since 2002.
3.3 Insights on a Future Potential Eruptive Scenario
For the 1970–1972 and 1994–1995 (Figure 3c) periods of fast increase of the elevation of the lava lake, the time of failure (tf) was estimated in the range (95% confidence interval for the linear regression) of December 1975 to January 1977 and December 1998 to December 2005, respectively. These periods fully overlap the effective date of the lateral flank eruption and subsequent lava lake drainage on January 1977 and January 2002. If we apply this method to the current increase of the elevation of the lava lake since 2016 (Figure 3a), we get a potential failure of the system in the March 2024 to November 2027 interval (95% confidence interval for the linear regression), centered on July 2025 (Figure 3c). Such a long-time delay between the precursory signs of FFM rupture and the occurrence of the latter (7–8 years in our case) is not so unusual in volcanic environments (Kilburn, 2018). This may be directly linked to the highly nonlinear nature of rock failure that can make the build-up to eruptions last several years (Chouet, 1996; Sparks, 2003).
The application of the FFM to forecast volcanic eruptions is classically associated to rock failure (Boué et al., 2016; Kilburn, 2018; Kilburn & Voight, 1998; Vasseur et al., 2017). Despite existing correlations between seismic signals and lava lake level fluctuations reported at Nyiragongo (Barrière et al., 2018), we are aware that linking the elevation of the lava lake to rock failure is not straightforward. If the pressure is expected to reach a plateau in the years preceding flank eruptions (i.e., 1972–1977, 1995–2002, and in the next months to years; Figures 2e and 3a), the induced stress is not expected to relax. In this situation, any magmatic or tectonic perturbation of the system could lead to rock fracturing and the propagation of a dike possibly leading to a flank eruption, as observed in 1977 and 2002. The active rifting system could be another trigger to the Nyiragongo external eruptive activity. Such a tectonic link has previously been evoked in the N–S trending fault system leading to the lateral 1977 and 2002 fissure eruptions (Carn, 2002/2003; d'Oreye et al., 2011; Komorowski et al., 2002; Poland, 2006; Tedesco et al., 2007; Wafula, 2011; Wood et al., 2017). A similar dependency on tectonics is also observed at the Erta'Ale volcano where the activity of the lava lake responds to regional magmatic-tectonic activity, rendering subtle perturbations at depth visible more rapidly at the surface (Barnie et al., 2016).
4 Conclusions
The forming of the new spatter cone in February 2016 within the summit crater of Nyiragongo volcano has accelerated the elevation rate of the crater floor (P3). This mirrors a pressure increase within the shallow magmatic reservoir that feeds both the lava lake and the spatter cone, both of which contributing to lava piling over P3. For the first time, a comparison with previous periods of activity preceding the two historical events of lava lake drainage (1977 and 2002) has been attempted, showing remarkable similarities with the ongoing activity. Since 2016, the activity is in a phase of fast pressurization of the magma system that could reach a plateau in a few months or years according to numerical simulations, which has successfully accounted for the evolution of the lava lake level over 30 years. At this time, the crater would contain a minimum volume of 16 Mm3 of molten lava. Whatever the magmatic and/or the tectonic trigger of the lava lake drainage at Nyiragongo may be, we stress that the eruptive activity could soon enter into a phase where the equilibrium of the system is precarious and where any perturbation of the system could disrupt it. If the current eruptive activity mirrors that observed for the last two cycles of activity, a potential failure of the system in the time between March 2024 and November 2027 could be a possible scenario. Field observations and numerical simulations presented in this study raise major civil protection issues for the upcoming years. We strongly encourage the international community and local authorities to be vigilant to the threat of any tectonic/magmatic perturbation that could break the equilibrium in the next years.
Acknowledgments
This mission could not be done without the invitation of the Goma Volcano Observatory (OVG), their logistic help, and their information provided on the activity of Nyiragongo volcano. MONUSCO is warmly thanked for providing a crucial support in several field deployments by helicopter in areas of unrest that are highly insecure. The ICCN provided logistics inside the National Parc of Virunga and provided full assistance with their very professional rangers in the field. Marc Caillet, Loïc and Melody Favre, and Johann Perrigault are thanked for their priceless support in the logistics of the descent into the crater. We also wish to thank Raphaël Paris, Hervé Sthioul, and Fabien Cruchon for their contributions in the 3D photogrammetry. We also warmly thank Stephan Kolzenburg for his helpful comments on an early version of the manuscript; the two reviewers, Matthew Patrick and Sébastien Valade for their constructive comments; and Nicole Frazier for her help with language editing. We acknowledge the INGV initiative LakeCarb and the French government IDEX-ISITE initiative 16-IDEX-0001 (CAP 20-25).
Open Research
Data Availability Statement
The data presented in this study are available online (https://doi.org/10.26037/yareta:4jx4pbkweneg5icpnf6bmxthra).