Banca de DEFESA: SANDRO GIOVANI DE FARIAS ALVES GOMES

Uma banca de DEFESA de DOUTORADO foi cadastrada pelo programa.
STUDENT : SANDRO GIOVANI DE FARIAS ALVES GOMES
DATE: 30/09/2026
TIME: 14:30
LOCAL: Videoconferência
TITLE:

Mining Seismicity in the Jacobina Region, Bahia, Brazil: Characterization, Source  Mechanisms and Uncertainties in Moment Tensor Inversion

KEY WORDS:

mining seismicity; moment-tensor inversion; structured empirical contamination; double-
difference relocation; Jacobina.

PAGES: 155
BIG AREA: Ciências Exatas e da Terra
AREA: Geociências
SUBÁREA: Geofísica
SPECIALTY: Sismologia
SUMMARY:

Seismicity associated with underground mining reflects the rock-mass response to excavation-driven stress redistribution and its interaction with pre-existing geological structures. This thesis investigates seismicity in the Jacobina underground gold mining district, Bahia, Brazil, by integrating two complementary studies. The first evaluates, in a controlled synthetic setting, how moment-tensor inversion (MTI) responds to different forms of waveform contamination; the second characterizes the observed seismicity, refines the relative geometry of the main clusters, and interprets source mechanisms through an explicit reliability hierarchy. In the synthetic study, band-limited Gaussian contamination, structured empirical contamination derived from small local events, and phase-randomized versions of the same empirical contaminants were compared for two observed six-station geometries and a combined 12-station configuration over source depths of 0.1–5 km. Full-waveform inversion (FW-I) was compared with P–S segmented inversion (PS-I). Gaussian and phase-randomized contamination show similar ensemble-level behavior, whereas structured empirical contamination selectively increases the occurrence of large errors, particularly under FW-I. For Network 2 at the highest contamination level, 32.1% of structured-minus-phase-randomized differences exceed +5◦ , compared with 7.3% below −5◦ ; at ±10◦ , the corresponding proportions are 16.4% and 2.6%. PS-I and increased station coverage reduce these error tails, while tests with independent empirical contaminants show that the depth of maximum susceptibility is not universal. In the observational study, 373 matched-filter detections yielded 320 located events between February 2022 and May 2023, with −0.5 ≤ ≤ ML 2.6. For a completeness magnitude = 0.7, 166 events yield = 1.14 ± 0.09; the rate above Mc b Mc decreases by approximately a factor of three near the end of July 2022 without a statistically resolved change in . Double-difference relocation retained 191 events, including 149 in the João Belo cluster. b Among 15 events analyzed by MTI, the hierarchical assessment produced one Class A, nine Class B, and five Class C solutions; eight Class A/B João Belo mechanisms are compatible with reverse to reverse-oblique kinematics. MTI centroid depths tend to be shallower than HypoDD hypocenters, but the offsets are treated as model-consistency diagnostics rather than relocation corrections. Exact non-double-couple fractions remain unresolved. Integration of relative cluster geometry, reliability-weighted mechanisms, mapped structures, and remotely sensed lineaments supports interpretation of João Belo as a structurally complex volume associated with a fault system or deformation zone. The central contribution of the thesis is to demonstrate that uncertainty in sparse local networks must be treated by source attribute: orientation, depth, and moment-tensor decomposition do not have equivalent resolution.

COMMITTEE MEMBERS:
Presidente - 1451214 - ADERSON FARIAS DO NASCIMENTO
Externo à Instituição - MARCELO BELENTANI DE BIANCHI - USP
Externo à Instituição - MARCELO PERES ROCHA - USP
Interno - 1298588 - RENATO RAMOS DA SILVA DANTAS
Externa à Instituição - SUSANNE TAINA RAMALHO MACIEL - UnB