VOLCANIC HAZARDS ASSESSMENT OF OLDOINYO LENGAI IN A DATA SCARCITY CONTEXT (TANZANIA)* AVALIAÇÃO DOS RISCOS VULCÂNICOS DO OLDOINYO LENGAI NUM CONTEXTO DE ESCASSEZ DE DADOS (TANZANIA)

The objective of our study is to establish an assessment of four volcanic hazards in a country threatened by the eruption of the OlDoinyo Lengai volcano. The last major eruption dates back to 2007-2008 but stronger activity in 2019 has revived the memory of volcanic threats to the Maasai and Bantu communities and human activities (agro-pastoral and tourism). The methods chosen have had to be adapted to the scarce and incomplete data. The volcanic hazards and their probability of occurrence were analysed on the basis of data available in the scientific literature and were supplemented by two field missions combining geography and hydro-geomorphology. Our study enabled us to map the hazards of ash fall, lava flows, lahars and avalanches of debris. Each hazard was spatialised by being ascribed an intensity. They are sometimes synchronous with the eruption sometimes they occur several months or years after a volcanic eruption. The results are the first step towards developing a volcanic risk management strategy, especially for the pastoral communities living around Lengai and for the growing tourist activities in this area.


Introduction
Volcanoes are complex geological systems that can produce a wide variety of dangerous phenomena during and after eruptions. These include glowing clouds, lava flows, pyroclastic flows, debris avalanches, ballistic projections, plumes and ashfalls, as well as volcanic earthquakes, landslides, gas emissions, floods, and fires (Baxter and Horwell, 2015). In response to these volcanic threats, states are attempting to put in place disaster risk management strategies despite existing uncertainties. To manage uncertainty, it is necessary to identify hazards, their intensity and spatial extent. The recognition of the issues at stake and the characterisation of territorial vulnerabilities make it possible to produce risk maps.
The difficulty therefore lies in acquiring reliable data to characterise hazards and volcanic risks. According to the available data, there are many methods for assessing and mapping volcanic hazards (Thouret et al., 2000;Leone and Lesales, 2009;Connor et al., 2015), such as the probabilistic method with the development of event trees (Newhall and Hoblit, 2002), and the use of digital models (Felpeto et al., 2007;Favalli et al., 2011;Tarakada, 2017). These methods require reliable data, acquired through the work of geologists, geomorphologists, volcanologists and geographers. Consequently, for poorly documented volcanoes, probabilistic scenarios are based on field investigations and, if possible, on modelling (Neri et al., 2013).
The objective of this article is to establish a diagnosis of the volcanic risks on the territory of Lengai. In 2008, 65,000 people were affected by the impacts of the eruption of the Oldoinyo Lengai (OLD), and several thousand people left their villages for a few weeks to several months (NEMC, 2008). NGOs provided food aid to almost 36,000 people (Msami, 2007). Faced with these poorly documented and insufficiently mapped volcanic threats, we present our progress in assessing the impacts of four volcanic hazards: ash fallout, lava flows, lahars and debris avalanches. Their assessment is based on probabilistic and deterministic approaches. The combined approaches underline the difficulties in scoping volcanic hazards for a region where data are scarce. Finally, we discuss the results and challenges we have faced using these approaches.

Study area
Ol Doinyo Lengai (OLD), "mountain of god" is venerated by the Maasai. For this ethnic community, the volcano represents the home of their god Ngai. Living near the OLD is not without its dangers and constraints but they are accepted. It's a territory made of assets and opportunities and a territory marked by a strong identity and cultural traditions (myths and rytuals).

Context
OLD is a stratovolcano in the vast East African Rift Plain, located in northern Tanzania, 16 km south of Lake Natron  Photo 1 -Lengai plain occupied by traditional settlements and the "mountain of god", January 2020 (Photography by T. Rey) Fot. 1 -A planície de Lengai ocupada por povoados tradicionais e a "montanha de deus", Janeiro 2020 (Fotografia de T. Rey) The OLD is highly studied for its petrological to emit natrocarbonatite lava (Bell et al., 1998;Dawson, 1998;Zaitsev and Keller, 2006;Kervyn et al., 2008a).
The eruptive activity of OLD is characterised by effusive eruptive phases of natrocarbonatite lava and explosive phases of nephelinite, known to have occurred in 1917in , 1940in -1941in and 1966in -1967in and the most recent in 2007Keller et al., 2010).
The understanding of eruptive activity in volcanic history has been reviewed by several international and Tanzanian scientific teams Keller, 2002;Keller and Klaudius, 2003). Seismicity is also the subject of particular attention with important instrumentation installed around the volcano (Albaric et al., 2010;Stamp et al., 2016;Weinstein et al., 2017). Seismic activity is currently monitored in real time by Ardhi University in Tanzania and the Institute of Geoscience and Mineral Resources in South Korea and is the subject of further studies on its "plumbing" and seismic-volcanic links with the Gelai.
The last major eruption took place in September 2007 and lasted until April 2008 (photo 2 and 3). It has been the topic of several publications (Mitchell and Dawson, 2007;Keller et al., 2010;Kervyn et al., 2010;Mattsson and Reusser, 2010) and many different mechanisms have been proposed to explain the activity associated with explosive episodes (Vaughan et al., 2008;Kervyn et al., 2008b). This eruption is the most well documented in terms of eruptive dynamics (Kervyn et al., 2008a) impacts of ash fallout on vegetation and resilience (De Schutter et al., 2015), the characteristics of ash deposition (Mitchell andDawson, 2017, Bosshard-Stadline et al., 2017), the influence of the release of fluorine contained in the 2008 lavas and ashes and its toxicity to water. The 2007-2008 eruption provides recent data on the impacts of volcanic hazards. However the Lengai space has also been the subject of other older volcanic phenomena, such as gravity movements (Kervyn et al., 2008b) and pyroclastic flows. These geological and hydro-geomorphological legacies have been grouped together in the geological map of Lengai (Sherrod et al., 2013) and have helped us to assess volcanic hazards.   Weather Station). These heavy rain cause floods and flash floods in urban area (Mikova et al., 2016). The Lengai Plain is also punctuated by hills resulting from debris avalanches identified by Kervin (2008) and lava flow deposits near the southern shore of Lake Natron (Sherrod et al., 2013).
The population and socio-economic activities within the Lengai area have vulnerabilities that expose them to varying intensities to volcanic threats and hydrometeorological hazards.

Population and activities
The OLD area, which includes the districts of Ngorongoro and Longido, had a population of nearly 300,000 (National census, 2012). Rural people accounted for nearly 97% of the population in these two districts, while urban people represented less than 3%, or nearly 17,000 people. However, these figures are now obsolete in view of the rapid urban growth that is driven by

Methods and data
Consequences of an OLD eruption the OLD are ash fallout, lava flows, ballistic projections, gas, lahars and debris avalanches, with direct and indirect effects.
All these hazards are not necessarily synchronous with the eruptive crisis. They have intrinsic characteristics, controlled by other factors such as rainfall, topography, gravity and earthquakes. They can then manifest themselves on temporalities that go beyond the eruptive crisis. We have chosen to evaluate 4 of the most frequent and intense hazards in the OLD area. The plurality of methods is justified by the scarcity of available data.

Ash fallout hazard
The analysis of ash fallout and deposit thickness were done using data from the 2008 eruption ( Using Tephra 2, we generated 6 probabilistic scenarios: a standard eruptive scenario (analogous to 2007-2008), then increased to 25% and 50%, with dominant E--S-E winds and without dominant winds.
Fot. 4 -Amostragem de cinzas vulcânicas, situação em janeiro 2020 (Fotografia de T. Rey, 2020).  With the Flow Order tool in ArcGis, we generated the catchment areas and hierarchised the hydrographic network according to Strahler's method. The sometimes inaccurate or incomplete data required a manual verification of the hydrographic network ( fig. 6). We then extracted data on the shape, area, slope and density of drainage (Table I). In the absence of measuring instruments and video recordings of lahars (and floods) in this area, we used the Manning-Strickler formula to estimate the flows and flow velocities. This approach combines field data with mathematical calculations, is an interesting solution for extrapolating hydrological dynamics during high-energy events (Rey et al., 2017).
The data needed for the calculations were taken from topometric measurements made in the main talwegs during the January 2020.

Lava flows hazard
The characterisation of lava flow hazard was based on the study of 72 reports between 01/1983 and 09/2020 (https://volcano.si.edu/volcano.cfm?vn=222120) and the work of Kervyn et al. (2008). We built up a database on the signs of volcanic activity of the OLD: lava flows (width, length, direction), evolution of hornitos, fractures on the cone, fumaroles, ash plume (Table II). The data made it possible to qualify the preferential directions of the lava flows ( fig. 7). The mapping of the lava flow hazard is also based on old flows identified by Sherrod et al. (2013).

Debris avalanches hazard
Four debris avalanches have been identified by Keller and Klaudius (2003): the largest event (4.9 km 3 ) named Zebra is dated around 10,000 years BP, the most recent Cheetah (0.2 km 3 ) around 2700 years BP (Klaudius and Keller 2006), and between these two events there is Orix (0.1 km 3 ) which is undated and a fourth debris avalanche identified on the shore of Lake Natron and dated at 793 ± 63 ka (Sherrod et al., 2013). All of these events correspond to flank collapses of the OLD (fig. 8).
The mapping of this hazard is therefore based on geomorphological legacies (Table III), the energy line method (Hungr et al., 2005) and recent geomorphological data acquired during the January 2020 field investigation.

Results
All scenarios are based on activity focused on the current Oldoinyo Lengai crater ( fig. 9).

Ash fallout
The results for the "no wind scenarios" highlight two cm. What allowed us to have realistic data was the integration of the granulometric parameters. Indeed, even with a significant increase in the eruption mass the scattering of ashes is relatively limited due to the coarse size of the ashes. According to the concentric model, the ash can extend more than 20 km. Engaruka and Engare Sero (more than 13,000 inhabitants) appear to be the most exposed to this hazard.
The results for the "scenarios with dominant E-S-E In the 25% and 50% scenario, the ash could be deposited more than 40 and 55 km from the crater. Therefore, in a maximum probability scenario, we estimate the danger zone to be 55 km from the crater.      ter et al., 2015). For buildings, we can incorporate thresholds to qualify the level of risk of roof collapse, such as a 10 cm threshold for dry ash (Blond, 1984).

Pyroclastic flows hazard (lahars)
On Photo 5 -Pyroclastic sediments observed at the bottom of the river, January 2020 (Photography by T. Rey, 2021).
Photo 6 -Observation of numerous flood sequences (and lahars) involved in the aggradation of the plain, January 2020 (Photography by T. Rey, 2021). We continue the analysis of lahars using the LaharZ model (Iverson et al., 1998) and our 6 diffusion and ash thickness scenarios (Tephra2 scenarios). The comparison of models and field data will make it possible to improve the spatial representation of this hazard.

Lava flows hazard
Our approach has made it possible to qualify two levels of hazard, a high and a moderate hazard. This distinction is based on the one hand, on the lava flows mapped by Sherrod et al. (2013)

Debris avalanche hazard
Four debris avalanches have been identified in the OLD area. Three debris avalanches were triggered on the northern flank. The deposits of the most energetic phenomenon were observed nearly 20 km from the present crater (Keller and Klaudius, 2003). The fourth debris avalanche was triggered on the eastern flank and propagated towards the E-N-E.
According to Kervyn (2008), the collapses occurred mainly at right angles to a dyke which would have generated underlying instability and fragility. The We have established a "high hazard" zoning for this probable debris avalanches, which is based on energy analogous to the Orix and Cheetah debris avalanches.
The debris avalanche scenarios do not take into account possible tsunamis on Lake Natron that could be generated by the instability of the northern flank of the OLD, nor do they take into account the possibility of the river being blocked by a debris avalanche from a gravitational movement on the western flank, a situation that would have already occurred during the Cheetah event from the E-NE flank to the Kerimasi and Gelai volcanoes (Klaudius and Keller 2006).

Discussion
The results underline the volcanic threats that affect the populations in this area. The methods chosen alternate between field data, historical and bibliographical data, and modelling. These choices have been oriented to compensate for the insufficiency or absence of data.
The results thus have limitations, but these cannot hide the singularity and originality of the results. These have made it possible to produce the first maps of volcanic hazards in the OLD space.
The scarcity of data is a problem encountered by other researchers in many territories, hence the need to adapt methods (Maharani et al., 2016;Selva et al., 2019). Lahars can occur during the eruptive crisis or later after the eruption (De Belizal et al., 2013), with a recurrence of lahars over several years such as Pinatubo (Crittendden and Rodolfo, 2002) or Merapi (Bignami et al., 2013;Jenkins et al., 2013;Wibowo et al., 2015).
In The probability of debris avalanches is more reduced: four known phenomena in almost 800,000 years. Their triggering is due to internal factors (earthquakes) and exogenous factors (rainfall, slope and crater instability) that are difficult to measure (Meunier et al., 2008;Miyabuchi et al., 2015). Nevertheless, certain wide and deep fractures that can be observed on the steepest slopes of the volcano (above 40°) can help us to map areas marked by active instability. It is also known that the lava covering the unconsolidated pyroclastics favours instability of the flanks.
The fall of ashes and lava flow have a temporality that follows the eruption. However, the impacts of ash fallout can occur over longer periods of time, extend beyond the Lengai area and rapidly cause a series of local and regional damages: damage to housing, ash-covered roads reducing or blocking access, diversion of air traffic and reduction in tourist flows, respiratory difficulties for humans and animals, ingestion of ash by livestock, burnt crops, and depending on the composition of the ash (Ph, acidity, minerals), sometimes lasting pollution of the soil and fresh water (Witham et al., 2005 ;Horwell and Baxter, 2006 and Engaruka-Ngwesuku, people were displaced for more than 10 months (NEMC, 2008). NGOs and the Red Cross provided food aid to nearly 36,000 people (Msami, 2007). In Lesele, ashes prevented families from returning to their bomass, which was abandoned (Courtesy of J. Keller and J. Klaudius in https://volcano.si.edu). In addition, fluorine concentrations posed a threat to the water supply for livestock and local people (De Schutter et al., 2015;Bosshard-Stadline et al., 2017). Livestock keepers then had to move their livestock to find healthy pastures and unpolluted water. For thicker ash coverage, the formation of a new soil layer is required to establish new plants, leading to a much longer recovery period.
Humans are also exposed to fluoride, and over the years they report diseases such as rachitis.

Conclusion
Although the volcanic hazards in Lengai are real, the scientific literature on their quantification is scarce and relatively poor, even in the presence of a relatively rich literature on the geology of space. . This situation did not provide a solid basis for the quantification of risks in the future. To fill this gap and prepare the ground for future quantification of volcanic risk, we proposed scenarios for four volcanic hazards. The scenarios were developed using a deterministic and probabilistic approach. These approaches need to be developed further, for example through the acquisition of a high-resolution DTM to better define the kinematics of lava flows and lahars, and to establish a geochronological framework for the sequences of lahars and debris avalanches that have been identified but not dated. In order to establish multi-risk maps, we continue our assessment of socioeconomic issues and their vulnerability to volcanic threats. An analysis of the accessibility and vulnerability of road networks is also conducted in the framework of evacuation scenarios (during the rainy and dry seasons).
An analysis of the perception of the Maasai communities living around the volcano is also underway in order to identify the complex relationships that exist between the Lengai God's abode volcano and the Maasai. In the end, we will be able to propose volcanic risk management strategies adapted to the local context.

Acknowledgments
The authors would like to thank all the team members