jackknife_location¶

Science¶

Leave-one-station-out relocation. The tool locates the full referenced phase set, then relocates again with each station's phases held out in turn, on the same grid × depth_mode menu as relocate. The spread of those trials is a sensitivity measure, not a confidence interval around the truth.

max_shift_km is the largest epicentral jump relative to the reference. most_influential_station is the station whose removal moved the epicentre farthest. A trial that does not converge or that times out is kept with converged: false and the envelope warns partial_result. Jackknife already skips a set with fewer than four stations; that is a station floor, distinct from relocate's four-phase floor.

A large shift when one station is removed means the location leans on that station (often a close-in P, or the only station closing a gap). It does not mean the catalogue origin is a given number of kilometres from the true hypocentre.

References¶

Efron, B. (1982). The Jackknife, the Bootstrap and Other Resampling Plans. SIAM. doi:10.1137/1.9781611970319
Lomax, A., Michelini, A. and Curtis, A. (2014). Earthquake Location, Direct, Global-Search Methods. doi:10.1007/978-3-642-27737-5_150-2

Contract¶

Generated from the live registry (ToolSpec). Field names, types and defaults come from the input and payload models; they are not hand-typed.


Group	`solution`
Class	3
Tool version	`1.1.0`
Highest declared tier	E
Timeout	90 s
Budget key	`jackknife_location`
`readOnlyHint`	true
`idempotentHint`	false

Input¶

Field	Type	Required	Notes
`picks`	object	yes	The phase set a solution tool works on.
`picks.base`	object (kind `catalog` \| `none` \| `solution`)	yes	discriminated on `kind`
`picks.phase_set`	string ∈ {`all`, `p_only`}	yes
`picks.exclude_pick_ids`	array[integer]	no
`picks.add_picks`	array[object]	no
`picks.add_picks[].tool_call_id`	string	yes
`picks.add_picks[].pick_index`	integer	yes
`grid`	string ∈ {`shallow`, `deep`}	no	default `"deep"`
`depth_mode`	string ∈ {`free`, `fixed_5`, `fixed_10`, `fixed_15`}	no	default `"free"`
`hypothesis`	string \| null	no	default `null`

Payload¶

Field	Type	Required	Notes
`reference`	object	yes
`reference.origin_time_utc`	string	yes
`reference.latitude_deg`	number	yes
`reference.longitude_deg`	number	yes
`reference.depth_km`	number	yes
`reference.depth_fixed`	boolean	yes
`reference.rms_s`	number \| null	yes
`reference.gap_deg`	number \| null	yes
`reference.num_phases`	integer	yes
`reference.num_stations`	integer	yes
`reference.min_distance_km`	number \| null	yes
`reference.uncertainty`	object \| null	yes
`reference.uncertainty.horizontal_km`	number \| null	yes
`reference.uncertainty.min_horizontal_km`	number \| null	yes
`reference.uncertainty.max_horizontal_km`	number \| null	yes
`reference.uncertainty.azimuth_max_horizontal_deg`	number \| null	yes
`reference.uncertainty.depth_km`	number \| null	yes
`reference.uncertainty.ellipsoid`	object \| null	yes
`reference.uncertainty.ellipsoid.len1_km`	number \| null	yes
`reference.uncertainty.ellipsoid.az1_deg`	number \| null	yes
`reference.uncertainty.ellipsoid.dip1_deg`	number \| null	yes
`reference.uncertainty.ellipsoid.len2_km`	number \| null	yes
`reference.uncertainty.ellipsoid.az2_deg`	number \| null	yes
`reference.uncertainty.ellipsoid.dip2_deg`	number \| null	yes
`reference.uncertainty.ellipsoid.len3_km`	number \| null	yes
`reference.quality`	object	yes
`reference.quality.converged`	boolean	yes
`reference.quality.grid_boundary_hit`	boolean	yes
`reference.quality.grid`	string	yes
`reference.quality.grid_floor_km`	number	yes
`trials`	array[object]	yes
`trials[]`	object	no
`max_shift_km`	number \| null	yes
`most_influential_station`	string \| null	yes
`n_trials`	integer	yes
`n_failed`	integer	yes
`config`	object \| null	no	default `null`
`config.grid`	string	yes
`config.depth_mode`	string	yes
`config.phase_set`	string	yes
`config.template_digest`	string \| null	no	default `null`

Example request¶

{
  "picks": {
    "base": {
      "kind": "catalog",
      "event_id": 1001
    },
    "phase_set": "all"
  },
  "grid": "deep",
  "depth_mode": "free"
}

Example payload¶

{
  "reference": {
    "origin_time_utc": "2024-06-15T08:12:03.410Z",
    "latitude_deg": 47.152,
    "longitude_deg": 24.497,
    "depth_km": 12.4,
    "depth_fixed": false,
    "rms_s": 0.24,
    "gap_deg": 108.0,
    "num_phases": 9,
    "num_stations": 6,
    "min_distance_km": 18.3,
    "uncertainty": null,
    "quality": {
      "converged": true,
      "grid_boundary_hit": false,
      "grid": "deep",
      "grid_floor_km": 250.0
    }
  },
  "trials": [
    {
      "station_removed": "XX.STA1",
      "n_phases": 8,
      "latitude_deg": 47.161,
      "longitude_deg": 24.51,
      "depth_km": 12.0,
      "rms_s": 0.27,
      "shift_km": 1.4,
      "depth_shift_km": -0.4,
      "converged": true
    }
  ],
  "max_shift_km": 6.2,
  "most_influential_station": "YY.STA3",
  "n_trials": 6,
  "n_failed": 0,
  "config": {
    "grid": "deep",
    "depth_mode": "free",
    "phase_set": "all",
    "template_digest": "sha256:0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef"
  }
}

What this tool does not tell you¶

What this tool does not tell you: absolute accuracy. Spread measures sensitivity to individual stations, not distance from the truth.

Warnings named for this tool¶

Code	When
`partial_result`	an individual trial timed out (`converged: false` on that trial)