Study Guide Type 3 — AI Agent Integration: Volcanology & Seismology Items 50–100

What it extracts

• Agent Design 1 — Real-Time Eruption Monitor: Specify the activation order, polling frequency, data schema, and alert threshold for an agent chaining KBT-GVP-HOLOCENE (baseline) → KBT-WOVODAT-MONITORING (unrest comparison) → KBT-USGS-VHP (alert level) → KBT-SPARKS-MONITORING (multi-parameter integration). What is the minimum dataset needed before the agent can output a probability?
• Agent Design 2 — Tephra Hazard Model: Design an agent that activates KBT-SPARKS-VOLCANICPLUMES (column height input) → KBT-FISHER-PYROCLASTIC (deposit classification) → KBT-NEWHALL-VEI (VEI assignment) → KBT-LOCKWOOD-VOLCANOGLOBAL (hazard zone delineation). Specify the input data requirements, intermediate outputs at each node, and final structured hazard map schema.
• Agent Design 3 — Probabilistic Eruption Forecast Engine: Build a Bayesian eruption forecasting agent using KBT-SIMKIN-VOLCANOSWORLD and KBT-LAMEVE-DATABASE as prior sources, KBT-MARZOCCHI-PROBFORECAST as the inference engine, and KBT-WOVODAT-MONITORING as the likelihood updater. Define prior construction, likelihood function, posterior output format, and uncertainty bounds.
• Agent Design 4 — Magma System Characterization: An agent receives whole-rock geochemistry and mineral analyses from a newly active volcano. Activate KBT-MCBIRNEY-IGNPETRO (magma genesis) → KBT-PUTIRKA-THERMOBAR (storage P-T) → KBT-CASHMAN-HOWVOLCWORK (eruption style threshold) → KBT-GOTTSMANN-CALDERA (if caldera system). Output: magma composition class, storage depth, explosivity index, and recommended monitoring priority.
• Agent Design 5 — Earthquake Rapid Response: An Mw 7.5 triggers the agent. It must chain KBT-USGS-COMCAT (aftershock retrieval) → KBT-GCMT-MOMENTTENSOR (fault mechanism) → KBT-GEM-GLOBALRISK (PSHA exposure) → KBT-IRIS-DMC (waveform retrieval for depth check). Define latency requirements at each step, fallback if a service is unavailable, and structured output for emergency managers within 10 minutes.
• Agent Design 6 — Climate Impact Chain: Eruption column height and SO₂ mass are inputs. Chain KBT-NEWHALL-VEI → KBT-SPARKS-VOLCANICPLUMES (stratospheric injection height) → KBT-OPPENHEIMER-ERUPTIONS (aerosol optical depth and cooling estimate) → KBT-MARTI-VOLCENVIRON (food security and ecosystem cascade). Specify unit conversions at each handoff and the confidence interval to carry through the chain.
• Agent Design 7 — Historical Seismicity Background Agent: For any location query, the agent must assemble a 500-year seismicity background using KBT-ISC-BULLETIN, KBT-GUTENBERG-SEISMEARTH (b-value fit), KBT-LAWSON-1906SFEQ (pre-instrumental methodology), and KBT-RICHTER-ELEMSEISMO (magnitude conversion). Define completeness correction logic and the output hazard rate table format.
• Agent Design 8 — Volcano Observatory Dashboard: Design an agent that synthesizes daily inputs from KBT-GVP-HOLOCENE, KBT-WOVODAT-MONITORING, KBT-USGS-VHP, KBT-JMA-JAPAN, and KBT-INGV-ITALY into a unified global volcanic activity dashboard. Specify data normalization across different alert level systems, conflict resolution when agency alerts differ, and output schema for a 24-hour situational awareness brief.
• Agent Design 9 — Stratigraphy-to-Hazard Pipeline: A field geologist uploads an outcrop description with 6 pyroclastic units. The agent activates KBT-CAS-VOLCANICSUC (facies analysis) → KBT-FISHER-PYROCLASTIC (genetic classification) → KBT-NEWHALL-VEI (historical VEI assignment from deposit) → KBT-SIMKIN-VOLCANOSWORLD (catalog cross-check). Output: eruption history reconstruction with confidence levels and recommended monitoring implications.
• Agent Design 10 — Geothermal Risk Screener: For any proposed geothermal development site, the agent queries KBT-WOHLETZ-GEOTHERMAL (resource and phreatic risk), KBT-DZURISIN-VOLCANODEFORM (baseline deformation requirement), KBT-GVP-HOLOCENE (volcano eruption history), and KBT-LAMEVE-DATABASE (large eruption recurrence). Output: go/no-go recommendation with risk tier, monitoring pre-conditions, and 50-year hazard context.
• Agent Design 11 — Dome Eruption Crisis Agent: Triggered when dome extrusion is confirmed at any monitored volcano. Activation chain: KBT-DRUITT-SOUFRIERE (dome collapse PDC model) → KBT-MCNUTT-VOLCSEISMO (hybrid swarm threshold) → KBT-MARZOCCHI-PROBFORECAST (collapse probability window) → KBT-USGS-VHP (alert level mapping). Specify the tilt-seismic data inputs, PDC run-out calculation, and exclusion zone boundary update cadence.
• Agent Design 12 — Cross-Source Conflict Resolution: An agent receives conflicting alert levels: GVP shows 'unrest', WOVOdat shows 'non-eruptive analogue', USGS VHP shows 'Advisory'. Apply KBT-MARZOCCHI-PROBFORECAST uncertainty framework to reconcile the signals. How does the agent weight each source, combine probabilities, and report the resulting alert with explicit uncertainty to its human operator?

Output Schema Fields