Essential Search Mathematics for SAR Managers & Planners Presented by Dan O’Connor NEWSAR.

Slides:



Advertisements
Similar presentations
Mathematics Review GySgt Hill. Mathematics Review Overview.
Advertisements

Introduction Creating equations from context is important since most real-world scenarios do not involve the equations being given. An equation is a mathematical.
POC, POD, POS Minnesota Wing Air Branch Director Course.
Paul James Doherty Park Ranger / GIS Specialist / Graduate Student.
LSP 121 Week 2 Intro to Statistics and SPSS/PASW.
LSP 121 Intro to Statistics and SPSS. Statistics One of many definitions: The mathematics of collecting and analyzing data to draw conclusions and make.
Statistical Analysis SC504/HS927 Spring Term 2008 Week 17 (25th January 2008): Analysing data.
Intro to Statistics for the Behavioral Sciences PSYC 1900 Lecture 3: Central Tendency And Dispersion.
Chapter 11 Data Descriptions and Probability Distributions
1 Confidence Interval for Population Mean The case when the population standard deviation is unknown (the more common case).
Monté Carlo Simulation MGS 3100 – Chapter 9. Simulation Defined A computer-based model used to run experiments on a real system.  Typically done on a.
Week 7: Means, SDs & z-scores problem sheet (answers)
Table of Contents Unit 1- Understand the Problem Unit 2- Gather Information Unit 3-Develop Solutions Unit 4-Implement a Solution Unit 5-Test and Evaluate.
Basic Statistics Standard Scores and the Normal Distribution.
Targeted Review of Major Concepts
Measurements of Central Tendency. Statistics vs Parameters Statistic: A characteristic or measure obtained by using the data values from a sample. Parameter:
REPRESENTATION OF DATA.
Objectives 1.2 Describing distributions with numbers
Dr. Serhat Eren DESCRIPTIVE STATISTICS FOR GROUPED DATA If there were 30 observations of weekly sales then you had all 30 numbers available to you.
Barnett/Ziegler/Byleen Finite Mathematics 11e1 Learning Objectives for Section 11.2 Measures of Central Tendency The student will be able to calculate.
1 DATA DESCRIPTION. 2 Units l Unit: entity we are studying, subject if human being l Each unit/subject has certain parameters, e.g., a student (subject)
Copyright © 2005 Brooks/Cole, a division of Thomson Learning, Inc. 1 Chapter 4 Numerical Methods for Describing Data.
© 2008 Brooks/Cole, a division of Thomson Learning, Inc. 1 Chapter 4 Numerical Methods for Describing Data.
Chapter 9 – 1 Chapter 6: The Normal Distribution Properties of the Normal Distribution Shapes of Normal Distributions Standard (Z) Scores The Standard.
Intro to Statistics and SPSS. Mean (average) Median – the middle score (even number of scores or odd number of scores) Percent Rank (percentile) – calculates.
An Introduction to Programming and Algorithms. Course Objectives A basic understanding of engineering problem solving process. A basic understanding of.
Analysis of Algorithms
© 2008 McGraw-Hill Higher Education The Statistical Imagination Chapter 6. Probability Theory and the Normal Probability Distribution.
Descriptive Statistics: Numerical Methods
Statistics Measures Chapter 15 Sections
Copyright © 2010, 2007, 2004 Pearson Education, Inc. Chapter 6 Normal Probability Distributions 6-1 Review and Preview 6-2 The Standard Normal.
 z – Score  Percentiles  Quartiles  A standardized value  A number of standard deviations a given value, x, is above or below the mean  z = (score.
3 common measures of dispersion or variability Range Range Variance Variance Standard Deviation Standard Deviation.
Essential Statistics Chapter 21 Describing Distributions with Numbers.
Central Tendency & Dispersion
IPS Chapter 1 © 2012 W.H. Freeman and Company  1.1: Displaying distributions with graphs  1.2: Describing distributions with numbers  1.3: Density Curves.
Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley Chapter 5 Describing Distributions Numerically.
Chapter 2 Describing Distributions with Numbers. Numerical Summaries u Center of the data –mean –median u Variation –range –quartiles (interquartile range)
1 Chapter 4 Numerical Methods for Describing Data.
Statistics What is statistics? Where are statistics used?
Chapter 2 Modeling Distributions of Data Objectives SWBAT: 1)Find and interpret the percentile of an individual value within a distribution of data. 2)Find.
Basic Statistics Six Sigma Foundations Continuous Improvement Training Six Sigma Foundations Continuous Improvement Training Six Sigma Simplicity.
Numerical descriptions of distributions
Variability Introduction to Statistics Chapter 4 Jan 22, 2009 Class #4.
Search Effort. The Balancing Act Limited Resources Limited Resources Planning Time in Search Area – how much time does it take to complete a search assignment?
Chapter 5 Describing Distributions Numerically Describing a Quantitative Variable using Percentiles Percentile –A given percent of the observations are.
The Normal Approximation for Data. History The normal curve was discovered by Abraham de Moivre around Around 1870, the Belgian mathematician Adolph.
Confidence Intervals. Point Estimate u A specific numerical value estimate of a parameter. u The best point estimate for the population mean is the sample.
MDFP Mathematics and Statistics 1. Univariate Data – Today’s Class 1.STATISTICS 2.Univariate (One Variable) Data 1.Definition 2.Mean, Median, Mode, Range.
Review Design of experiments, histograms, average and standard deviation, normal approximation, measurement error, and probability.
Chap 5-1 Discrete and Continuous Probability Distributions.
Week 2 Normal Distributions, Scatter Plots, Regression and Random.
Introduction Creating equations from context is important since most real-world scenarios do not involve the equations being given. An equation is a mathematical.
Mission Aircrew Course Search Planning and Coverage
BAE 5333 Applied Water Resources Statistics
Describing Distributions Numerically
The Statistical Imagination
Descriptive Statistics (Part 2)
Objective: Given a data set, compute measures of center and spread.
Reasoning in Psychology Using Statistics
Collecting and processing of information Presentation 4.5.1
Statistical Analysis with Excel
Introduction Creating equations from context is important since most real-world scenarios do not involve the equations being given. An equation is a mathematical.
Statistical Analysis with Excel
Statistics Variability
Basic Practice of Statistics - 3rd Edition
An examination of the purpose and techniques of inequality measurement
Essential Statistics Describing Distributions with Numbers
Basic Practice of Statistics - 3rd Edition
Basic Practice of Statistics - 3rd Edition
Presentation transcript:

Essential Search Mathematics for SAR Managers & Planners Presented by Dan O’Connor NEWSAR

“Windows” CASIE Computer-Aided Search Information Exchange FREE at

“Closed” System “Defective” Probability “Open” System No ROW 100% POC ROW + Segments = 100% POA IPP “Background” 3 Types of Search Systems 30% 50% SA Less Than 100% POA or POC Physical Limits Physical & Psychological Limits

1. Theoretical vs. Statistical Search Area (SA) What’s the difference?

THEORETICAL Search Area The Straight-Line Distance that a Lost Person could have traveled “in theory” over the Elapsed Time since reported Missing Rate x Time = Distance (as est. of radius) 2 mph x 12hrs = 24 miles A radius of 24 miles means a Circular Search Area of 1,810 Square Miles! Equivalent to a 40 mi by 45 mi Area!

STATISTICAL Search Area An AREA based on Distances that other Lost Persons have traveled in the PAST. Ideally, these distances traveled are compiled by Lost Person Category (child, elderly, hiker, etc.) Search Managers typically draw Statistical Search Areas based on the MEDIAN (50 th Percentile) & 75 th & 90 th & 95 th Percentiles Maybe should be called “Potential Search Area”

Q. Why are Potential Search Areas Drawn as Circles?

A. Because in the Absence of CLUES, we have no idea about the Lost Subject’s Direction of Travel

Sources for STATISTICAL Distances Traveled... 1.Ken Hill (Nova Scotia data) published in the NASAR MLPI Text & CASIE 2.“Lost Person Behavior,” Robert Koester 3.ISRID Koester & Twardy et al 4.SARSTATISTICS.org (under development) 5.Your OWN or other Local Agency Data

CASIE Source Distances Traveled

2. The MEDIAN: the value which divides the Data in Equal Halves. 50% is At or Above the Median And 50% is At or Below the Median “The Median home price in the area is $300,000.” Half sold at or above, half sold at or below.

The POSITION of the MEDIAN Is NOT the VALUE of the MEDIAN! IMPORTANT!

To find the POSITION of the MEDIAN in a SORTED Dataset use: MEDpos = 0.5 * (n+1) For 99 data points, the POSITION Of the Median = 0.5 * (99+1) = 50

17 SORTED Lost Person Distance Traveled Data Pts POSITION DATAPercentile (P) Position of 63.4Median 74.1formula * (N+1) th4.8Median Average: 67thPercentile (12/18) Sum 6.2Mean or "Average"

The MEDIAN is More Stable, The MEAN is More Variable 1.Consider our 17 Data Points, from 0.5mi to 20.3 mi with Mean=6.2 mi and MEDIAN=4.8 mi... 2.If we ADD 2 more data points at 1 mi and 30 mi, the Mean goes to 7.1 mi, but the MEDIAN=4.8! 3.The Mean is sensitive to Outliers – the Median is NOT!

The MEDIAN also defines the position of the 50 th Percentile Data: 0.5mi 4.8mi 20.3mi Percentile : The MEDIAN lives here at the 50 th Percentile OR end of the 5 th Decile

Why Use the MEDIAN? When to Use the MEDIAN? Why not 75 th or 90 th Pctile? What should “r” radius be? Questions on Radius “r”

IPP 20.3 mi 4.8 mi 50% AREA= ? Which Area is easiest to search? Both represent 50% of cases...

AREA of a Circle = pi * r ^ 2 For r = 4.8, Area = 3.14 * (4.8 * 4.8) = 72 sq units For r = 20.3, Area = 3.14 * (20.3 * 20.3) = 1294 sq units Area of Outer Circle (annulus) = 1294 – 72 = 1222 sq units

Area of an Annulus in CASIE

IPP 20.3 mi 4.8 mi 50% AREA= 72 sq mi AREA= 1222 sq mi Which Area is easiest to search? Both represent 50% of cases... 50%

IPP 20.3 mi 4.8 mi 50% pDen= 50% / 72 sq mi = 0.69% per sq mi pDen= 50% / 1222 sq mi = 0.041% per sq mi Another way to look at it... pDEN Probability Density: % Statistical POA per Unit Area

CONSENSUS POA is different from Statistical Probability. The Area with the top 50% of cases might be assigned only 10% POA initially as a Region NOTE!

RESOURCES Are LIMITED TIME Is Limited HIGH Coverage is Required Increased Urgency for Good Confinement It’s a Tradeoff for a smaller SA WHEN to Search Within the Median

Statistical Circles are NOT Limits to the Search Area... Go wherever the CLUES Lead!

“Closed” System “Defective” Probability “Open” System No ROW 100% POC ROW + Segments = 100% POA IPP “Background” 3 Types of Search Systems 30% 50% SA Less Than 100% POA or POC Physical Limits Physical & Psychological Limits

3. Analyzing OWN Agency Data A. Sort and Compute Percentiles B. Compute the “75% Plus” Range of Finds

Advantage to “75% Plus”... Uses STANDARD DEVIATION in Data to estimate Variability in LPDT values Very Robust for SMALL Datasets “Conservative” way to proceed

MED = 7 Sorted Data LP Distance Traveled 11 Data Points in Miles 75 th Percentile = 10 (9 th Position) MEDpos 0.5 * (11+1) = 0.5 * 12 = 6 The Data Value “7” is at the 6 th Position in the Dataset

For “75% Plus” Compute Sample STANDARD DEVIATION in Excel by using: +STDEV(data range) then for “75% Plus” range calculate: Mean – (2 * SD) = lower bound Mean + (2 * SD) = upper bound

Lower = 0.0 Sorted Data LP Distance Traveled For MEAN=7.63 & SD = Upper = % Plus Range = [Mean – 2*SD to Mean + 2*SD] Reflects VARIABILITY Within the Data; When Lower Bound is NEGATIVE, Use Zero

4. Methods for Creating a Consensus In CASIE there are 3 Methods available: 1. MATTSON (numeric POA’s = 100%) 2. O’CONNOR (use Verbal Cues) 3. PROPORTIONAL (rate relative to Baseline #)

MATTSON

O’CONNOR

PROPORTIONAL

Initial POA’s from Proportional Consensus

5. 2-Methods for Updating a Search Bayes Formula, With ROW OPOS Summation, Without ROW

Bayes Formula, With ROW Based on P(A|B) or “the Probability of A, Given B” The fact that I have searched in B affects the probability of finding the subject in A. Once B is searched, the POA of A goes UP. BA

Bayes Formula, With ROW BIG SCARY Formula... Hard to Do by Hand, especially multiple updates Do It In CASIE or a Spreadsheet!

Bayes Formula, With ROW Update in CASIE Seg#POA-0PODPOA-1 ROW27.50% % %86%6.59% % % % %

Overall Probability of Success, Without ROW Seg#POA-0PODPOSPOA % % 2 86%28.66%4.67% %86%28.66%4.67% OPOS0% %--

6. Optimizing Resources Brute Force, Calculate to Exhaustion (David Lovelock, Retired Math Prof, U of AZ) Washburn Algorithm (Alan Washburn, Naval Post-Graduate School) Both require estimating Resource POD

Optimizing Resources in CASIE go to top menu “What If” then “Resource Allocation Advice 2. Create a New Table

Resource Allocation Table: Estimated POD for Each Resource in Each Segment of Interest

WHY BRUTE FORCE?

BRUTE FORCE ADVICE – 3 Scenarios

Washburn Algorithm – 1 “Optimal” Scenario

7. The Mathematical Importance of CONFINEMENT At a 1 Mile Radius (5,280 feet), Step ONE FOOT farther and the AREA increases by 33,179 sq ft. About 3/4ths of a Football Field (210’ x 150’) to the 74 Yard Line!

8. COVERAGE & POD Use the Exponential Detection Function (EDF) to find POD from COVERAGE At COVERAGE = 1, POD = 63% “Efficient” At COVERAGE =2, POD = 86% “Thorough” Note: It takes TWICE as much Effort (Resources) to get a Coverage=2 as it does to get Coverage=1.

The “Expanded” EDF Too Efficient, Not Thorough Too Thorough, Not Efficient Optimal region 63% 86%

Determining Grid Spacing from Critical Separation Chart 1: Convert CS to est. ESW Chart 2: Select Desired Coverage Chart 3: Obtain Spacing Example: For a CS of 0.6  est ESW=48; for 86% POD  Coverage=2, & Spacing = 24. (Note: for AMDR, skip Chart 1; multiply AMDR by 1.5 to calculate est ESW, then use Charts 2 & 3) 86% 63% Version 1.2 Source:

9. Estimating EFFECTIVE SWEEP WIDTH (ESW) In the Absence of an Appropriate Detection Table, Sample the Terrain to be Searched using... CRITICAL SEPARATION, or Avg. Max. Detection Range (AMDR) and Adjust for an Estimate of ESW

Mt. Greylock base trail, Berkshires, MA – Various Seasons. Source: Rick Toman, MSP The Complexity of the Ever-Changing LandSAR Environment

Determining Critical Separation - 1 CS Under Prevailing Conditions If the Object changes, or the Conditions change, a new CS value must be computed! ½ CS

Determining Grid Spacing from Critical Separation Chart 1: Convert CS to est. ESW Chart 2: Select Desired Coverage Chart 3: Obtain Spacing Example: For a CS of 0.6  est ESW=48; for 86% POD  Coverage=2, & Spacing = 24. (Note: for AMDR, skip Chart 1; multiply AMDR by 1.5 to calculate est ESW, then use Charts 2 & 3) 86% 63% Version 1.2 Source:

10. K9 POD for SAR Managers Major Environmental Factors that Affect K9 POD 1.Sun Angle (High is Bad) 2.Wind (Still is Bad) 3.Cloud Cover (Clear is Bad)

10. K9 POD for SAR Managers You debrief a K9 team on a hot August day in Arkansas... They have been out for 4 hours between 10am and 2pm. The sky is clear and the wind is still. The Handler says that their POD=95% for 40 acres. Q. What is your Response to that POD?

BALONEY!

Many factors go into estimating K9 POD... Best bet... BUY The MLPI Text at the NASAR Bookstore and refer to the Table on p.225!

11. Calculating Cumulative POD 1.Table in MLPI & Field Guide 2.Exp Detection Function (EDF) 3.CASIE (different vs. same teams)

Determining Grid Spacing from Critical Separation Chart 1: Convert CS to est. ESW Chart 2: Select Desired Coverage Chart 3: Obtain Spacing Example: For a CS of 0.6  est ESW=48; for 86% POD  Coverage=2, & Spacing = 24. (Note: for AMDR, skip Chart 1; multiply AMDR by 1.5 to calculate est ESW, then use Charts 2 & 3) 86% 63% Version 1.2 Source:

12. GRID SEARCH PLANNING Formulas Assume Ground Searcher SPEED Of 3.5 Hours Per Mile... How Fast is that in mph? 1 Mile / 3.5 Hours/Mile = mph

12. Find Required # of Searchers

13. Find Searchable Area

14. Find Hours needed to search

15. Find required Spacing

Bonus! Coverage & Track Spacing from #15 Inputs

MLPI Planning Exercise (p.223) 1.High Pressure! Congressman’s Relative Lost! 2.IC wants 80% POD over 1 sq. mile 3.Gives you 100 Ground Searchers 4.ESW estimated to be 60 feet 5.How long will this take? You have 2 minutes!

MLPI Planning Exercise (p.223) 1.Solution: Use CASIE! 2.Find Coverage at 80% POD 3.Find Spacing at Coverage = 1.6 with ESW=60 4.Use HOURS Planning Formulas for Time 5.Answer: 5hrs (4.9 rounded up)

THANKS!

ENCORE? T-CARDS!