Piedade And Boa Viagem Recife 2026¶

This paired case study records a timeline-separated AI1SAD replay for two Greater Recife incidents in Pernambuco, Brazil. It is designed to show what the current model can and cannot represent without adding a Recife scoring retune.

The replay uses four runs:

Piedade strict pre-incident replay
Piedade quiet-day comparison
Boa Viagem strict pre-incident replay
Boa Viagem quiet-day comparison

Verified Incident Records¶

Field	Piedade Beach	Boa Viagem Beach
Location	Piedade Beach, Jaboatao dos Guararapes, Greater Recife, Pernambuco, Brazil	Boa Viagem Beach, Recife, Pernambuco, Brazil; Acaiaca / Padre Bernardino Pessoa corridor
Replay coordinates	`-8.171, -34.914` approximate operational point	`-8.126, -34.899` approximate operational point
Date	`2026-05-31`	`2026-06-01`
Approximate local time	`13:40`	`15:10`
Victim context	Child bather, age 11	Teenage / young-adult woman bather, age 19
Outcome	Serious injuries with left-leg amputation	Serious injuries with right-leg amputation
Source-attributed species assessment	CEMIT preliminary adult bull shark, Portuguese source term `tubarão-cabeça-chata`, approximately 2.5 m if retained from source	CEMIT preliminary adult tiger shark, Portuguese source term `tubarão-tigre`, approximately 3 m

The CEMIT species assessments are stored as preliminary, source-attributed classifications. They are not used as independently verified taxonomy in the strict pre-incident model inputs.

The paired replay does not generalize both incidents as bull-shark events. It keeps Piedade's source-attributed bull-shark context separate from Boa Viagem's source-attributed tiger-shark context and does not assume the same individual shark caused both incidents.

Replay Inputs¶

The strict pre-incident runs do not include hindsight signals. Tide, current, turbidity, rainfall, SST, and weather values are marked missing where unavailable.

Run	Timeline rule	Included signals	Excluded signals
Piedade strict pre-incident	Before `2026-05-31T13:40-03:00`	Bather/swimming activity and baseline reef-barrier/channel context	Boa Viagem incident, later response activity, post-incident species assessment as model input
Piedade quiet day	Same location, season, time-of-day	Same baseline activity and habitat context	Incident-specific signals
Boa Viagem strict pre-incident	Before `2026-06-01T15:10-03:00`	Bather/swimming activity, baseline reef-barrier/channel context, and Piedade as a known recent regional interaction	Post-Boa-Viagem response activity and post-incident species assessment as model input
Boa Viagem quiet day	Same location, season, time-of-day	Same baseline activity and habitat context	Piedade recent-interaction signal

Replay Results¶

Run	Warning Score	Activity Hazard Score	Surveillance Priority Score	Confidence	Interpretation
Piedade strict pre-incident	`0.0` low	`0` low	`9.0` low	`0.43`	Very weak pre-incident signal stack; available context only defines an observation corridor.
Piedade quiet day	`0.0` low	`0` low	`9.0` low	`0.43`	Same output because no incident-specific pre-signal is available.
Boa Viagem strict pre-incident	`0.0` low	`0` low	`23.0` low	`0.48`	Prior Piedade interaction adds a timeline-valid regional cluster signal, raising surveillance attention by `14.0`.
Boa Viagem quiet day	`0.0` low	`0` low	`9.0` low	`0.43`	Without the Piedade recent interaction, Boa Viagem returns to the same weak baseline.

Dominant Factors¶

Piedade strict pre-incident:

reef_dropoff_habitat_proximity: 5
activity_context: 4

Boa Viagem strict pre-incident:

recent_interactions_nearby: 14
reef_dropoff_habitat_proximity: 5
activity_context: 4

Species-specific explainability metadata:

piedade_source_attributed_bull_shark_suitability: 0 points; CEMIT preliminary tubarão-cabeça-chata context only
boa_viagem_source_attributed_tiger_shark_suitability: 0 points; CEMIT preliminary tubarão-tigre context only
species_agnostic_recent_interaction_lift: 14.0 points; shared corridor cluster context after Piedade, not a same-species or same-individual assumption

Missing Signal Sources¶

The replay explicitly marks these gaps:

live sighting ingestion
Recife/Pernambuco regional pack
tide and current observations
turbidity and water clarity
rainfall/runoff lookback
SST/anomaly
human exposure estimates
vessel/fishing context
telemetry or monitoring-program ingestion

Operational Recommendations¶

Recommended patterns:

baseline drone patrol windows:
morning
midday
late afternoon / dusk
adaptive surge monitoring after:
recent severe incident
nearby sighting
tagged-shark receiver detection later
high human exposure
tide/current overlap later
turbidity/runoff overlap later
shoreline_parallel_sweep
reef_barrier_parallel_sweep
reef_gap_focus_scan
channel_mouth_pass
post_sighting_focus_area
nearshore_channel_ladder_search
posted-swim-zone observation focus
future tide/current and turbidity-informed corridor once supported

The current model identifies the Boa Viagem cluster signal only after Piedade becomes timeline-valid input. It does not claim the Piedade strict pre-incident signal stack was strong. This result points to coverage gaps: live observations, Recife-specific habitat/water-state context, and human-exposure timing would matter more than tuning around these two events.

Artifacts¶

Caveat¶

This replay is a bounded operational-surveillance artifact for review and planning. It does not replace official beach, lifeguard, emergency, wildlife, or public-health guidance.