Transcript Document

Tracking Uncertainty in
Search and Rescue
Planning
Art Allen
.
U.S. Coast Guard
Office of Search and Rescue
[email protected]
The basic steps for
SAR Planning
1. Initial Conditions
.
2. Drift Predictions
3. Resource Allocation
4. Actual search effectiveness
5. Updating search conditions
Rescue
or
Suspend
Databases
Search Results
Build a
SAR Case
Assemble
Search
Plan
Disseminate
. Search
Plan
Environmental
Now &
Forecasts
Capture
Search
Results
Search
Plans
Results
Field
Initial Conditions
Uncertainties
1. Scenarios
• Subjective relative weights
.
between scenarios
• (e.g. 80%A or 20%B)
• Each scenarios will have its own
set of initial conditions
Initial Conditions
Uncertainties
1. Where did the incident occur?
2. When did the incident
occur?
.
3. What are we looking for?
4. How long will they survive?
Databases
Store Case Info.
Background
Search object
info.
User inputs
Capture
.
Initial Case
Data
(Input and weight
Scenarios)
Initial POC
Initial POSv
Initial States
.
.
.
.
.
.
.
.
.
.
.
.
Rescue
or
Suspend
Databases
Search Results
Build a
SAR Case
Assemble
Search
Plan
Disseminate
. Search
Plan
Environmental
Now &
Forecasts
Capture
Search
Results
Search
Plans
Results
Field
(2) Assemble Search Plan
A) Compute Probability Density Distributions
Databases (Case, SRU’s)
Initial POC
Initial POSv
Initial States
Drift/ Survival/State
of Replications
(Leeway / Drift Model)
(Survival Model)
(Detection Model)
Drifted POC
Updated POSv
Updated States
Search Results
Now and Forecast Winds,
Currents, Waves,
SST & Air T, Visibility, etc.
Probability
Distributions
1. Drift uncertainties
2. Survival uncertainties
.
3. Detection uncertainties
Drift
Uncertainties
1. Surface current errors & dispersion (u’,v’)
2. Target uncertainty .
• What are we looking for?
3. Leeway uncertainty
4. Wind and Waves
Surface Currents
1. Sources:
• Historical ship-drift global data sets
• Global, regional &. coastal models
• Direct on-scene measurements
2. Dispersion
• Random walk – 0th order Markov model
Self-Locating Datum Marker Buoys
th CODE
SLDMBs
– Air-deployable
• Random
flight – 1st7/10
order
Markov
drifter, GPS positions & SST via Argos
model
SLDMBs
.
Self-Locating Datum Marker Buoys Airdeployable 7/10th CODE drifter, GPS
positions & SST via Argos
Surface currents from
CODAR
.
CODAR / SLDMBs
•Black: Actual
SLDMB
Trajectory
•Red: Trajectory
Predicted From
NOAA Tidal
station
•Blue: Trajectory
Predicted From
CODAR Data
.
CODAR / SLDMBs
•Black: Actual
SLDMB
Trajectory
•Blue: Trajectory
Predicted From
CODAR Data
•Red: Trajectory
Predicted From
STPS predictions
.
Leeway classes
1. Statistical analysis of field experiments:
• Indirect and Direct
. methods
• US, Canadian, Japanese & Korean
2. Leeway taxonomy by Allen and Plourde
(1999) combined results into 63 categories.
.
Approximations
Wind speed and object drift is approximately
linearly related
Approximations
Different objects drift differently
Undrogued life raft
Life raft with drogue
Objects do not drift exactly downwind!
Constrained or Unconstrained Linear Equations
Used in Monte Carlo Simulations
Winds and Waves
1. Global, regional & coastal models
2. Direct on-scene measurements
.
• Both wind and waves act through leeway
Probability
Distributions
1. Drift uncertainties
2. Survival uncertainties
.
3. Detection uncertainties
Survival Modeling
1. UK Immersion statistics
2. CESM Hypothermia model
Mathematical model
.
Physiological data
Survival Uncertainties
1. CESM / UK statistics limited to cold
water survival
2. Warm water survival factors:
1. Still hypothermic
. (loss of heat)
2. Dehydration (lost of water)
3. Sleeplessness (lost of sleep /
restoration)
4. Fatigue (loss of available energy)
5. Circadian rhythms (cycles in all
above)
6. Predation (loss of blood)
Probability
Distributions
1. Drift uncertainties
2. Survival uncertainties
.
3. Detection uncertainties
Detection
Uncertainties
Lateral range curves & sweep widths based
upon limited field tests and MSPP
.
• Limited environmental conditions
• Limited Target sets
• Limited SRU / sensor combinations
Rescue
or
Suspend
Databases
Search Results
Build a
SAR Case
Assemble
Search
Plan
Disseminate
. Search
Plan
Environmental
Now &
Forecasts
Capture
Search
Results
Search
Plans
Results
Field
(2) Assemble Search Plan
B) Capture Resources
Databases
Background
SRU info.
Users inputs
Capture SRU
Availability
SRUs called
(Position, Status
Capabilities, Limitations)
SRU Risks
Winds,Waves,
SST & Air T,
Visibility, etc.
(2) Assemble Search Plan
C) Allocate Resources
Databases
Drifted POC
Updated POSv
Updated States
SRUs called
1) Suggest Optimal Survivor
Search Plan
2) Modify Plan to suit
3) Capture Plan to execute
Forecasted Winds,Waves,
SST, Air T, &
Sensor Environmentals
Search
PLANS
(3) Disseminate Search Plans
Search
PLANS
Support various
output methods
to disseminate
search plans.
To Shore
To Vessels
To Aircraft
(4) Capture Search Results
Databases
USCG Vessels
USCG Aircraft
Other searchers
Capture description of
completed searches
including both positive
and negative results.
Search Results
Nowcasted/Observed Winds,Waves,
SST, Air T, &
Sensor Environmentals
Rescue
or
Suspend