During clean up of an early-warning-radar facility in the early 60’s, a sled-load of small arms rounds in ammunition boxes was reportedly dumped into the northern end of Troutman Lake. Troutman Lake is a shallow fresh-water lake just south of Gambell, on the Northwest Cape of Saint Lawrence Island, Alaska (figure 1). Due to the arctic climate of region, thick ice covers the lake until mid-June. Taking advantage of the climate, a TEM survey using sled-mounted equipment was conducted in May 2000, while the ice was still 1.5 m thick, but most of the snow cover had melted and 18 hours of daylight extended the potential workday.
Although the original sled-load of stacked ammunition boxes was a large conductive mass, other UXO characteristics are possible. Large breakers churn across Troutman Lake during fall storms, so the ammunition may now be in separate ammunition boxes scattered across the lake bottom, or even dispersed as individual small-arms rounds. The unusual survey environment and range of possible UXO characteristics controlled both the TEM equipment configuration and survey parameters.
Thick ice cover made it easy to tow sled-mounted equipment across the ice (figure 2). Averaged TEM transients were recorded every 3 seconds using an in-loop TEM configuration. A real- time, kinematic-phase differential GPS unit kept track of the sled’s position as it was towed along line by the equipment operator and recorded positions accurate to better than 0.1 m every 5 seconds.
Spatial Sampling Requirements
The northern end of Troutman Lake is fairly shallow, 2 to 3 m deep across the center of the search area. As the half-width of an in-loop TEM anomaly is proportional to target depth (Nabighian and Macnae, 1987), a sampling interval of 2 m along line generated two to three anomalous points over detectable targets (figure 3). A 7.6 m (25 foot) line spacing was used over the entire survey area, with 3.8 m (12.5 foot) line increments across the more crucial central search area (figure 4).
Background-geology TEM responses have smooth spatial variation, while UXO TEM anomalies are compact positive peaks. The contrast allows suppression of background-geologic responses by spatial filtering.
Time Sampling Requirements
Transient data were recorded at 26 delay times spaced logarithmically between 7 and 570 microseconds (usec). Recording a complete transient waveform provides several advantages. Compact metallic conductors produce a TEM signal dominated by an exponential decay, exp(- t/τ), at later transient delay times (Kaufman, 1978). Target size and conductivity control the time constant (τ). Larger and more conductive objects have larger characteristic time constants. Recording full transients saves sufficient information for time-constant estimates, returning useful information about UXO properties.
UXO exponential time constants control the optimal transient delay time for target detection. The exponential-decay signal from UXO is masked by a geologic-background TEM response which decays in proportion to t-k, where k = 5/2 for a uniform half-space and can vary between 3/2 and 7/2 over layered-earth backgrounds (McCracken, et al, 1986) (figure 5). UXO signals are most detectable when the target/background response ratio is at its maximum, which occurs at delay
time =k ⋅ τ . Different size search objects have different time constants, consequently no single
transient delay time is optimal in searching for UXO with a wide range of sizes.
Optimal TEM loop size is controlled by expected UXO depth. Small 1 by 1 m loops are optimal for the detection of shallow objects. However, the 1 to 3 m depth to UXO on the bottom of Troutman Lake increases the optimal loop size. Plotting peak anomaly amplitude versus TEM loop size (figure 6) shows a relationship between UXO depth and optimal loop size. For UXO
0.5 m below the loop, peak anomaly amplitude reaches a maximum for 1.2 by 1.2 m loops. When the UXO depth is increased to 2 m, 5 by 5 m loops are optimal, although practical considerations put an upper limit on loop size. Our 4.6 by 4.6 m transmitter loop was near optimal size, and yet was small enough to construct as a light sled-mounted structure. Using a smaller 1.5 by 1.5 receiver loop improved the early-time characteristics of the TEM system at the cost of reduced late-time signal amplitude.
Using larger TEM loops does sacrifice sensitivity to near-surface targets. Figure 7 shows anomaly amplitude versus target depth for 1 by 1 m and 4.6 by 4.6 m loop TEM systems. Using larger loops reduces the system’s sensitivity to near-surface objects by a factor of 8 relative to a 1 by 1 m loop system, but a large loop system has relatively more sensitivity to objects at depths more than of 0.8 m.
A 2000 by 3400 foot area over the northern end of Troutman lake was mapped with 426,000 line-feet of profile data. Twenty-eight targets were selected using both profile and plan map data presentations. A plan map of 40 usec data (figure 8) shows narrow target anomalies superimposed on a smoothly varying geologic-background response. A profile view along a line through Anomaly C (figure 9) shows how the response of a strong target anomaly stands out from geologic background.
Analysis of a transient from Anomaly C (figure 10) shows how a large conductive object produces a strong late-time anomaly dominated by a single exponential decay. Exponential decay forms straight-line segments on log-linear plots of dBz/dt versus time. Line slope is proportional to the exponential time constant. The target signal from Anomaly C indicates a 410 usec time constant, consistent with a conductive object the size of a 55 gallon drum. In contrast, Anomaly B (figure 11), has a 20 to 80 usec target signal with a much shorter time constant, 14 usec, indicative of a smaller conductor.
No single equipment configuration is optimal for all situations. Survey results can be optimized by adjusting both equipment design and survey parameters. Larger loops improve TEM’s sensitivity to deeper objects, at the cost of portability and sensitivity to shallow UXO. The optimal time window for UXO detection in the presence of a background geologic response is proportional to the target’s dominant exponential time constant, a function of target size, shape and conductivity. Recording a full transient waveform allows UXO detection using more than one time window when searching for a range of UXO types. Analysis of full-transient data can produce information about the target’s size, shape and conductivity.