The (almost) t) No Di Dig Remedial Investi tigati tion 26 Febru - - PowerPoint PPT Presentation
The (almost) t) No Di Dig Remedial Investi tigati tion 26 Febru 26 February 2015 ary 2015 Steve Stacy, PG ARCADIS, U.S. e-mail: steve.stacy@arcadis-us.com Office Phone: 703-465-4234 Mobile Phone: 425-891-4507 Agend
e-mail: steve.stacy@arcadis-us.com Office Phone: 703-465-4234 Mobile Phone: 425-891-4507
2
3
4
5
EM61-MK2: 8.72 Acres MetalMapper: 3.44 acres
6
7
8
9
10
11
12
13
14
15
Advanced Classification Results Dig Results Cluster Number of Anomalies in Cluster Number of Anomalies Selected for Intrusive Investigation Suspected UXO Number of UXO Found Dig Results 1 4 1 Doesn't match library well Illum disk 2 4 1 Mortar Tail Boom 3 4 1 Frag 4 2 1 No Contact 5 3 1 Tail boom part 6 10 1 Tail boom part 7 7 1 Frag and fuze parts 8 11 3 60mm mortar tail booms 9 10 1 Fuze Part Fuze Parts 10 11 1 Fuze Part Tail boom part 11 99 7 Fuze Part 60mm tail booms and fins 12 14 6 60mm Mortar 60mm Illumination Bodies 13 15 2 Fuze Part 60mm and 81mm Mortar Parachute Assemblies 14 4 1 Hand Grenade Fuze shipping clip 15 6 2 Fuze Part 81mm Mortar parachute assembly and frag 16 10 3 81mm Mortar 1 81 mm M374 HE Mortar; 81mm illum body; scrap metal 17 13 8 60mm Mortar 4 4 60 mm HE M49 Mortar; Mortar tail boom part; 60mm Illum body; frag 18 3 1 81mm Mortar Drive Shaft 230 42 5
16
Target ID Decision Statistic UXA_UXO TYPE Dig Type Dig Result 318 0.9807 60mm M49A3 Mortar UXO 60 mm HE M49 Mortar 370 0.9564 MD Tail Boom Part 372 0.9483 UXO 60 mm HE M49 Mortar 236 0.9453 UXO 60 mm HE M49 Mortar 373 0.9427 UXO 60 mm HE M49 Mortar 118 0.9192 60mm M69 Practice Mortar MD 60mm Illumination Body 169 0.8627 NA MD Frag
17
18
Limited intrusive investigation
Limit impacts (e.g., T&E species) Reduce evacuations (e.g., residential, offices) Limited funding
Can determine nature and extent of MEC Sufficient to evaluate remedial alternative costs
No ROC curve – can’t fully evaluate performance AGC with more digs could better determine dig selection threshold Helps to have anticipated TOI BSIs
19
21
22
EM EMI sensor data ta from meta tallic
ts can be fit t with th dipole model Model paramete ters:
Object t Locati tion, Xo, Y , Yo, Z , Zo Di Dipole polarizati tions used to to identi tify
Given locati tion, model inversion is linear and fast t De Dete tecti tion Filte ter
Grid field with th Xo, Y , Yo locati tions (0.1m) Specify filte ter depth th, Zo (0.2m (0.2m) At t each locati tion, select t window of data ta (1.6x1.8m (1.6x1.8m) an ) and apply lin d apply linear in ear inversion ersion for polarizati tions Filte ter outp tput t is “goodness-of-fit” t” betw tween model and data ta at t th that t locati tion (coherence, 0.0 – – 1.0) Filte ter peaks indicate te object t locati tions
tional Threshold:
Model-bas based, m ed, min inim imum peak signal from small ISO at t maximum depth th of inte terest t
Pick all ll signa signal l pea eaks s above ve th this th threshold
ter Threshold: – Em Embed model-based signal from small ISO in signal-free regions of measured data ta – Apply dete tecti tion filte ter to to (Model+Noise) (Model+Noise) an and d look at t peak filte ter amplitu tude – Apply filte ter to to just t measured noise for SNR – Filte ter can dete tect t to to deeper depth ths th than signal alon alone e
De Dete tecti tion filte ter may increase number of dete tecti tions over simp simple le peak eak signal signal (imp (impro roved ved SNR SNR) ) Use inverte ted polarizati tions to to pre-screen locati tions 1,2 and 3-dipole inversion at t filte ter peak (X (Xo,Y ,Yo) ) to to handle multi tiple objects ts at t or near one locati tion - if inversion produces additi tional sources >0.4m from original filte ter peak repeat t inversion using data ta cente tered on new source locati tions Resulti ting sources are examined and culled based on size, decay and amplitu tude metr trics to to only sources th that t could be a 37m 37mm or larg
er Fit t locati tions from th the inversions used as th the final locati tions for th the cued ta target t list t
+ - Final Detection ○ - Initial filter peak Using the dipole filter Detection process reduced final target list from 134 amplitude based anomalies to 13 dipole filter anomalies