Contribution de la méthode neuronale à l’optimisation de investigations de géophysique au Cameroun.

Abstract

In this study, the contribution of the neural method to the optimization of geophysical data processing in Cameroon is highlighted through a densification technique. For this purpose, the Lom-Pangar region in Eastern Cameroon was chosen as a target area because it has one of the lowest coverage of terrestrial geophysical data in Cameroon. Starting from 466 points of insitu measurements, we obtain 960 points after densification. The results obtained after densification are considered satisfactory from a geostatistical point of view, in particular, statistical (good correlation R2 =0.9811, low error values: RMSE= 0.0804 and MBE= 0.0003) and gravimetric with firstly a strong similarity between the Bouguer obtained from the neural method and that obtained exclusively from in-situ data, then a gradient zone that would be similar to an identity related to the CCS zone. Densified data analysis highlights three major directions in the Lom-Pangar area (NE-SW, WNW-ESE, and ENE-WSW), all of which are identified as belonging to the Pan-African domain and corresponding to three geological domains of different age and evolution, namely the syn-tectonic granites in the east, the Lom series in the center, and the metamorphic series in the northwest. The ENE-WSW orientation appears as the main direction. (2) Residual regional separation associated with 2.5D modeling characterize the Lom-Pangar area as belonging to the broad corridor of negative and positive anomalies. The positive anomalies to the south and north are associated with intrusions of dense material, emplaced by upwelling of heavy material from the lower crust or of mantle origin. The top of the basement constituting the maximum depth of the basin is at 17.5 km. The models obtained show that the basement of the region is granite-gneissic. Migmatites and gneisses provide the fault contacts in the Lom-Pangar area. (3) New faults have been identified and considered as lateral and vertical extensions of the Sanaga Shear Zone (SSZ) or the Central Cameroon Shear Zone (CCSZ). New structural layers were clearly identified in the southern Sembé and western Pangar localities, reflecting the strong dynamics and geological complexity of the study area. These findings more than corroborate those obtained previously in the region. They thus contribute to validate the neural method as a method for optimizing geophysical data.