pit house question

Can we use total field magnetics to find buried pit houses
beneath layers of volcanic ash?
Visible pit houses at Bridge River, B.C. (Prentiss et al., 2009)
Can we use total field magnetics to find buried pit houses
beneath layers of volcanic ash?
Approach, constraints and thoughts:
• The site, Anangula in the Aleutian Islands, has up to
two meters of volcanic ash burying some of the pit
Bridge River Study:
vertical magnetic gradient
The vertical magnetic gradient
observations from Prentiss et
al. (2009) provide an analog to
model similar features in the
Aleutians. The result helps
confirm the likely success of
magnetic exploration to map
the layout of the site. Note that
the observed values of vertical
magnetic gradient are very
high with a range of about 200
nt/m. A portion of the
amplitude of the circular highs
(red rings) is probably due to
the excavated soil.
Vertical gradient magnetometer observations from the Bridge
River, British Columbia site provide an analog for modeling
Use forward magnetic models of truncated vertical cylinders
To test a worst case situation produce a signal much less than
that of the Bridge River results
Use forward magnetic models of truncated vertical cylinders
Signal range is about 25% of that for the Bridge River vertical
gradient observations
Total Field Intensity (TMI) and Vertical Gradient of the Total Field:
these anomalies are calculated over simulated pit houses
Pit houses modeled as vertical cylinders 2m tall, with radius = 6m, and tops at the surface.
Amplitudes of anomalies are scaled so that the range of the vertical gradient is about 25% of the
anomalies from Bridge River. This tests a weaker case.
Some measure the vertical gradient in the field but it can be calculated from total field observations
as one type of edge detector (aka image enhancement).
Recalculated TMI and Vertical Gradient Anomalies:
simulated pit houses are now 2 meters below the surface
The range of the vertical gradient is now reduced to about 10 nt, or 5% of the values measured at
Bridge River. This is due to lowering the top of the modeled pit houses to 2m below the surface.
The anomalies are still very strong.
Contaminate the weak forward signal with substantial noise:
determine if we can still recover signal
Simulated Fractal Noise: Amplitude ~Equal to the TMI Signal
signal/noise ~ 1 !
Random (1/f2) magnetization (A/m)
simulating very variable soil properties
Magnetic field (TMI) calculated from
the 1/f2 distribution of magnetization
Adding this magnetic field to that for the two meter deep pit houses will simulate a
‘bad’ field situation. The noise (+/-20 nt) is a bit greater than the TMI’s range (+/-15nt).
Add the Fractal Noise to the Signal From 2m Deep Pit Houses:
a field situation with an extremely low signal to noise ratio
The general location of the deep
pit houses, even though greatly
obscured by the noise, is still
apparent in the total field anomaly
even without any processing.
The location of the pit houses is very
difficult to distinguish in the vertical
gradient anomaly. This is because the
vertical gradient is a high pass frequency
operator and biases against the signal
from the deep pit houses.
Post-processing Enhances the TMI Anomalies
Deep equivalent magnetic layer after
separating shallow source components
with matched filtering
Maxima of the horizontal gradient of
the pseudogravity transform of the
TMI; highs estimate edges of sources.
Even in this low signal/noise situation the pit houses standout.
Pit house model parameters for a pretty bad (if not worse) case field scenario:
House diameter = 6 meters; depth to top = 2 meters
Surface magnetic anomalies scaled to 25% of Bridge River vertical gradients
After burial to two meters signal amplitude is about 5% of Bridge River values
Two meter layer of fractal noise added to signal to simulate volcanic soils
Model signal to noise ratio ~ 1
1 meter spacing of observations; they would be at 0.1 meter along lines in the field.
The modeled situation is quite pessimistic relative to
known results at Bridge River. Regardless, the pit
houses are still very apparent.
If the Bridge River results are at all characteristic, the
situation at Anangula in the Aleutian Islands could
easily be much better than this model.

similar documents