like a 3- to 4-ft-long, lO-in.-diam shell would be). These type of items-could very 

 easily be situated so that a convergence of negative magnetic gradients would be 

 immediately (2 to 3 m) north of the convergence of magnetic positive gradients. 



Target Location and Analysis by Maximum 

 Likelihood Estimation Method 



Arete" Engineering Technologies Corporation (AETC) of Arlington, VA, 

 examined and conducted additional post-processing of about 60 percent of the 

 magnetometer data from Sea Bright. AETC used a target characterization 

 procedure based on matching measured magnetic anomalies to magnetic dipole 

 fields using Maximum Likelihood Estimation techniques. They inferred the object 

 size from the dipole moment using an empirical relationship (Bell, DeProspo, and 

 Prouty 1996). 



Arete" Engineering pointed out that there was a significant range of magnetic 

 response from the UXO, even for items of fixed caliber, and the standard- 

 deviation about the mean correlation for similar-sized targets was about 

 25 percent. Some of the variability in apparent size for specific ordnance items 

 was due to remnant magnetization, but the primary factor was the shape and 

 orientation of objects on their magnetic signatures. When the long axis is aligned 

 with the earth's field, the induced dipole moment of such an object is much larger 

 than, for example, the dipole moment that is induced when the object is lying 

 transverse to the earth's field. This indicates that future calibration field tests of 

 the cesium magnetometers must be conducted with test objects lying both parallel 

 with and perpendicular to the earth's magnetic field. 



One hundred magnetic anomalies were selected from the survey data for 

 detailed analysis to demonstrate the target characterization procedures. The data 

 were taken from the six long north-south lines. A histogram of the distribution of 

 anomaly strengths is shown in Figure 28. With few exceptions, the apparent 

 dipoles were oriented more or less to the north, suggesting that ordnance in this 

 area is lying on the seafloor approximately parallel to the New Jersey shore. 

 Distribution of the estimated cross-track locations of the 100 anomalies is shown 

 in Figure 29. Positive values are to the right of the survey track line, and the 

 shaded area shows the detection swath width for the magnetometer array used 

 during this survey. With the magnetic sensor array flying at about 1.7 m above 

 the seafloor, the system detected objects at a range slightly over 4 m to either 

 side. Sensitivity studies based on dipole anomalies embedded in uncorrelated 

 Gaussian noise demonstrated that for these ranges, typical ordnance can be located 

 with 10- to 20-cm accuracy using the survey data. 



The distribution in depth for the test anomalies is shown in Figure 30. Most 

 objects were lying on the seafloor, but a small minority appeared to be hovering 

 50 to 100 cm above the bottom. It is not clear if these peculiar results were due 

 to faults with the altitude sensor, raised seafloor areas, or some other undetected 

 problems. Possibly they represent long or irregular-shaped marine debris float are 



26 



Chapter 6 Magnetometer 



