slot size. A filter sock with a mesh of 120- to 150-
signed circuit board (Fig. 9b). In addition, a pres-
m covers the outside of the screen (Fig. 9). After
sure transducer (Druck PDCR-35D) is also in-
we installed the wells, each hole was allowed to
stalled in each well and hourly fluctuations in the
collapse around the screen as a way to maintain
water level (piezometric surface) are recorded.
minimally disturbed conditions. Monitoring wells
Data stored within the logger are subsequently
drilled through permafrost were screened at a
downloaded into a portable computer for analy-
depth beginning 3 m below the bottom of the
sis and graphical display.
permafrost. During drilling through permafrost,
As mentioned, flow sensors are calibrated
we followed standard procedures to stop ground
within a flume. For this study it was filled with a
water from the suprapermafrost aquifer from
commercially available sand mixture (sieve no.
flowing down into the hole.
812) that is generally representative of materials
in each well screen interval. This calibration pro-
vides an empirical relationship between the sen-
Automated ground water
sor readings and seepage velocity, as well as di-
flow system
CRREL's prototype ground water flow system
rection. A more precise and accurate velocity
uses the thermal tag and trace technique (e.g.,
measurement can be obtained by calibrating each
Chapman and Robinson 1962, Hess 1982) to mea-
sensor to the actual materials surrounding each
sure flow direction and seepage velocity in the
well screen. The calibration used in this study
monitoring wells in the Canol Road area (Fig. 8).
produces a relative velocity that is considered
accurate to within 0.6 ft/day (0.15 m/day).
This system employs a sensor that closely follows
the configuration of a commercial probe, which
had limited applicability to permafrost areas. Ex-
tensive work has developed a more refined sys-
ANALYSIS AND
tem for accurately measuring the direction and
INTERPRETATION
magnitude of slowly flowing ground water (Will-
iams et al. 1995).
We have developed preliminary maps of the
The prototype system uses a sensor with four
permafrost distribution and hydrogeology of the
pairs of platinum Resistive Temperature Devices
Canol Road area that are based upon interpreta-
(RTDs) in steel tynes that are arrayed in a circular
tion of the GPR records, borehole logs, terrain
pattern around a point source heater (Fig. 9a).
analyses and related data (e.g., Lawson et al. 1993,
Heat is transferred preferentially by ground wa-
Ecology and Environment 1994) on the surficial
ter flow from the heater to the RTD's. Conductive
geology and ground water conditions.
heat transfer occurs through glass beads in which
the RTD tynes are imbedded.
Permafrost distribution
Temperature fluctuations between paired RTDs
Permanently frozen and unfrozen sediments
relative to background levels are recorded. This
in the Canol Road area are complexly distributed,
temperature differential is then used to calculate
with highly variable annual depths of thaw and
the direction of flow and seepage velocity. Each
depths to the bottom of permafrost. The spatial
sensor should be calibrated in a flume filled with
distribution of deep and shallow unfrozen zones,
materials representative of those surrounding each
including the active layer above permafrost, are
particular well screen to obtain absolute velocity
depicted in Figure 10a, while a second map (Fig.
measurements. A constant displacement pump is
10b) portrays the occurrence of unfrozen sedi-
used to regulate seepage rates from 1 to 10 ft/day
ments beneath permafrost. Both maps show the
(0.3 to 3 m/day).
locations of zones that are completely unfrozen
Flow sensors are suspended from steel rods
but surrounded by permafrost.
lowered and locked into position in each well
Deep thaw zones--typically less than 3 to over
after orientation to magnetic north. The suspen-
15 m deep--are commonly associated with dis-
sion system was designed at CRREL to ensure
turbed terrain (e.g., roads and clearings), while
minimum slack in connections and precise orien-
shallower active layers of less than 3-m depth
tation of the sensor at depth (Williams et al. 1995).
characterize undisturbed permafrost terrain (Fig.
The direction and velocity of seepage are mea-
10a). Areas with thin active layers (1.0 m or less)
sured hourly, with system functions controlled
are commonly associated with thick permafrost
by a Campbell CR10XT data logger. Signals are
that extends into bedrock. Areas or zones that are
transmitted and received through a CRREL-de-
unfrozen from the surface to bedrock are com-
9
Previous Page
