Seismic Hazard Levels in Guinea, West Africa

The occurrence of seismic activities in Guinea is infrequent, although located in a stable continental region. The hazard assessment level in Guinea, West Africa was determined by a probabilistic approach for 10 sites across the region. The calculation was carried out for 10%, 2% and 0.5% probability of exceedance in 50 years using a homogenized 100-year catalogue compiled from different seismic sources; Three prediction relations, developed for Eastern and Central North America for the stable continental region; and the R-CRISIS program. The levels of hazard estimated were high in the Palaeozoic area of Guinea. A uniform b -value of 0.70 ± 0.12, and individual activity rate (λ) were calculated for the three seismic zones. The maximum PGA values estimated for the study region are 0.08 g, 0.27 g, and 0.57 g for 475, 2475 and 9975 years return periods, respectively. Finding from this study will be useful in planning for the regional infrastructure.


Introduction
The location of the Guinea region far away from any known active plate boundary. The Guinea region is not seismically known and the occurrence of moderate to large earthquakes are infrequent based on available records. The first reported earthquake in Guinea was in 1795, with an estimated surface magnitude of 5.2, which caused considerable damage in the city of Labé 1 . The 1818 earthquake with a 5.9 surface magnitude at the epicentre in Futa Djallon massif, northern Guinea. A strong earthquake with considerable damage and a surface magnitude of 4.0 occurred in the Kakulima region but no damage was reported. There was a strong shock in the western part of Guinea, where some local houses along the Konkouré river were destroyed in April 1928 with a surface magnitude of 4.8; renewed shocks followed this in August that caused landslide 1 . Since then, some historical earthquakes have also been reported in 1935 and 1939, respectively. Finally, on 22nd December 1983, north-western Guinea experienced a strong earthquake of 6.3 magnitude (Mw). The epicentre of the event located in Gaoual, close to the border of Guinea-Bissau. It resulted in approximately 10 km of surface rupture, which extensively damaged buildings, killing over 300 people and more than 4,000 houses were destroyed 2,3 .
The seismic hazard across the cities of Guinea, West Africa has not been assessed. The seismic source model developed for this study region is separated into three zones, using the available seismicity and local geological features. The ground motion model is constructed using three different strong-motion attenuation equations.
Finally, the hazard estimates for ten cities are computed at 475, 2475, and 9975 years return periods to be the first to be made available in the literature for the Guinea region. As such, they provide valuable data for risk assessment and mitigation, land use management, and planning for present and future infrastructures across the selected cities. Guinea earthquake recorded in the study. Rocks in Guinea is affected by Rokelide Orogeny like the one in Sierra Leone, characterised by Archean and Pan-African structures 5 .

Compiled Earthquake Catalogue
A carefully compiled list of earthquake catalogues is a critical input in all seismic hazard assessments regardless of the approach adopted. The historical data sources used for this study was compiled 1  where ML = local magnitude (ML), Ms =surface magnitude, and mb = body magnitude.
Since different catalogues reported the same events, duplicates were manually removed using the criteria 9 , which include events located within 10 km and recorded within a period of 2 minutes. For duplicate events, the order of priority is as follows: ISC earthquake catalogue followed by the USGS-NEIC earthquake catalogue, and finally 1 . The complete catalogue was declustered to remove foreshocks and aftershocks, and seismic swarms, using an algorithm 10 , implemented 11 . The declustered catalogue contained 27 events in the moment magnitude range 4 ≤ Mw ≤ 6.3. The spatial and temporal distributions of the declustered catalogue (Fig. 2).  Since information regarding the local faults for this study region is poorly documented, the determination of seismic hazard was defined on the seismicity and geological setting of the study area. Three seismic zones used to estimate the hazard levels for the ten sites are shown in (Fig. 5). Zone A represents the Palaeozoic craton, while zone B and zone C are characteristic of the Archaean and Lower Proterozoic rocks. Calculated recurrence parameters of b-values and lambda (λ) for each zone expressed exponential using Gutenberg-Richter distribution 16 , where N = number of events with magnitudes equal to, or greater than, M, a-value = activity rate which defines Gutenberg-Richter relation intercept at M equal zero.
The b-value indicates the relative number of large and small earthquakes and representing slope of the Gutenberg-Richter relation, λ =b ln(10).

Figure 3. Catalogue completeness periods
It is common to use a unique b-value for source zones in low-to-moderate seismicity due to limited recorded data 17,18 . As a result, a uniform b-value was calculated and adopted for all the zones (Table 2) and Figure 4.
The a-value calculated is provided in Table 2. The λ-parameter is known to vary significantly from the different zones within a given area. It was estimated differently for each zone by taking the average number of earthquakes for magnitude equal to or higher than the minimum magnitude (Mmin). The maximum-likelihood method 11 18,23,24 . The accuracy of the focal depths is generally poor in this study region owing to the limited available information. However, based on the seismic sources available for this study, the earthquake foci are within 0 to 15 km. Consequently, the hazard analyses were carried using the same value of 10 km for all zones. The ground-motion prediction equations (GMPEs) are essential in the definition of the ground motion model. Although regional attention relationships calibrated from local seismic data are preferred, these were not available for the study region owing to limited data, and in three zones, lack of recorded strong-ground motions. Consequently, three Ground motion prediction equations (GMPEs) [25][26][27] developed on Eastern North America that is frequently used for a stable continental region where the study region is applicable estimated the likely ground motion in this study. Three GMPEs were used for the hazard calculation by assigning the same weight for each of them, thereby representing confidence levels in the model.

Seismic hazard calculations
The seismic hazard calculations 12,13 in this study region were performed using the R-CRISIS software A probabilistic seismic hazard analyses software developed with support from II-UNAM, the Instituto de Ingeniería at UNAM, México 19 . The software is available for free on the R-CRCIS website upon request and was used extensively in the assessment. The analysis is carried out on 0.5-degree grids under the following conditions: • Calculations were made for 0.5%, 2%, and 10% probability of exceedance in 50 years, corresponding to return 475, 2475 and 9975 years return periods, respectively.
• Minimum magnitude of Mmin = 4.0 was adopted, considered to cause engineering damage.
• Calculations were carried out to produce PGA in terms of hazard maps, hazard curve and seismic hazard curves In Guinea West Africa, where information on faults is not available, the input parameter such as Mmax values was assessed and considered in the hazard calculations. Mmax was assessed by adding an arbitrary value of 0.5 22 .  Figure 7 plots the seismic hazard curves for ten selected cities within the Guinea region (PGA values listed in Table 3). For 475 years return period (Fig. 6a) corresponding to 10% probability of exceedance in 50 years, hazard levels close to  Figure 7a shows the seismic hazard curve in terms of Peak spectral acceleration for the ten sites. The gap in (Fig. 7a) confirms the cities with higher levels of seismic hazards. Figure 7   The maximum credible magnitude (Mmax) determined based on the large observed earthquake magnitude in the region may be underestimated. No strong ground motion was available in this study region. Therefore, the attenuation relations adopted are based on numerical simulations. The compiled catalogue covered 100 years duration with few earthquakes and was used to estimate recurrence parameters which may introduce error in the analysis. Three Seismic source zones are proposed based on geology and seismicity due to lack of information on local faults. The study produces hazard maps for ten sites using a probabilistic approach. The results show that at 475, 2475 and 9975 years return periods, there are significant seismic hazards within the Conakry and five other cities close to the Gulf of Guinea -despite low to moderate levels of seismicity reported in the region.

Hazard results and Discussion
From this study, six out of ten cities selected (Gaoual, Labé, Kindia, Kamar, Manéah and Conakry) fall within the high hazard levels for the new seismic hazard maps. The hazard levels estimated are critical since the ten cities now seeing sizeable infrastructure development. Therefore, the seismic risk associated with an increase in population and the expansion of the regional infrastructure can be significant. Thus, to protect the local population and to sustain the region's economic development, it is critical to mitigating against such risks, which, from the engineering design view, can be achieved by designing and building new infrastructure using the seismic hazard maps proposed in this study.

Data availability
The dataset generated during and/or analysed during the current study are available from the corresponding author upon request.