Comparing Geophysical Methods for Determining the Thickness of Arctic Sea Ice: Is There a Correlation Between Thickness and Surface Temperature? Ross Robertson, Logan Fisher, Katie Mankowski, and Rhett B Herman Several small geophysical surveys were conducted on the Chukchi Sea ice just offshore from the Naval Arctic Research Laboratory near Utqiagvik (né Barrow), Alaska, in March, 2016. The goal was to investigate a possible correlation between the surface temperature and the thickness of the sea ice, as well as to test a potential new method for more accurately determining ice thickness. In addition, this survey tested the efficacy of a low-cost sensor sled of our own design that could obtain this so-called “microclimate” data. The equipment used in this survey included the following: • an MLX90614 IR sensor to obtain surface temperatures • six DS18B20 temperature sensors to obtain air temperatures • ground penetrating radar, 500MHz and 1,000MHz • capacitively coupled resistivity array (“traditional” & “non-traditional” setups) • an ice drill The survey lines are pictured below. Image created using Google Earth. Introduction • The sled was designed and built at Radford University. • A wooden sled was used as a platform. • Data acquisition was controlled via an Arduino Mega microcontroller. • Data were stored on a microSD card using an OpenLog data logger. • A pvc pipe held an MLX90614 IR temperature sensor that pointed down and measured surface temperature with ±0.5° accuracy, and ±0.02° resolution. • Six Dallas Semiconductor DS18B20 temperature sensors with vertical spacing of ~20 cm collected ambient air temperatures with an accuracy of ± 0.5° with 12-bit resolution (±0.05° ). • Data were collected at ~88cm intervals using an odometer wheel on the sled of our own design. • A switch on the sled handle was used to put a mark in the data every 10m as a check on the accuracy of the odometer wheel. Sensor Sled Various geophysical methods were used to study the Arctic sea ice including electrical resistivity, ground penetrating radar, and measuring the ice surface temperature. The results have yielded the following conclusions: • The microcontroller-based sensor sled worked as expected in this prototype stage. The odometer wheel accurately measured distances as verified with measured surface marks. The MLX90614 IR sensor measured small changes in the surface temperature. The DS18B20 temperature sensors obtained air temperature data within 1.5m of the surface (see side mini-poster). • The surface temperature seems to be influenced by the makeup of the subsurface. This was observed when the resistivity data and IR surface temperature were compared along a line that started on the shore and ended more than 160m onto the sea ice. There was an abrupt change in surface temperature when going from the shore and grounded ice onto the sea ice. This type of surface temperature change was also observed when the surface temperatures were compared to the 900m GPR line (500MHz). • The 500MHz GPR antenna gave a better resolution for the ice/water boundary than the 1,000MHz antenna. This was observed when the two were compared along the same 200m line. • The resistivity array in the “traditional” setup was not effective in studying the sea ice. This was likely due to the thinness of the sea ice. The minimum size of the array had 2.5 meter dipoles. This directed most of the transmitter’s signal into the seawater with its low resistivity. It appears the array was incapable of measuring such low resistivity values using more than one receiver. Each receiver must acquire its own signal as well as forwarding “along the chain” the signals from the other receivers (those further from the control console) within the one second cycle time. A remedy may involve increasing the cycle time. However, the next item may be more relevant for dealing with this limitation. • Using the resistivity array in an “expanding dipole-dipole” configuration may yield the depth to the ice/water boundary. These data were analyzed using the cumulative resistivity method of Robinson, et al, [1988] and the results compared to the measured ice thickness. This procedure was performed twice, with the calculated and measured values for ice thickness being within 2% of each other. • The different slopes of the cumulative resistivity graphs are understandable in the context of the expected relative resistivities of the layers. For the first (top) layer the sum of the resistivities Σ simply adds the same resistivity – that of ice – to the same values coming before (i.e. in the shallower parts of the ice). Once the effective depth of investigation starts to include the lower resistivity of the seawater, the sum of the resistivities Σ now adds a lower resistivity value to the previous ones. This leads to an expected decrease in slope, which occurs on the graph. Assuming the expanding dipole-dipole array were to penetrate into yet another layer, then the slope would again be expected to change. The slope did decrease, suggesting that there is a lower resistivity to whatever lies beneath the seawater. This could be expected from an ocean floor having a number of substances dissolved in the saltwater slurry. • Future work: The apparent success of both the sensor sled and the expanding dipole-dipole array is guiding our future work. We are currently working to develop a microcontroller-based “micro resistivity array.” This array will have multiple (e.g. 24+) electrodes spaced much less than a meter apart, like a small version of a typical commercial galvanic resistivity array. The microcontroller will activate 4 electrodes (2 current, 2 voltage) at a time and switch between them automatically. The current source will be a dc-to-ac inverter capable of delivering several hundred Watts of power [see e.g. Herman, 2002]. Conclusions A number of ground penetrating radar (GPR) surveys were performed with 1,000MHz and 500 MHz antennae. The 500MHz provided the clearest image of the ice/water boundary. The two images below compare the two frequencies. The ice surface was almost completely flat due to the strong winds at the time scouring the surface. The top image shows the results of the 1,000MHz antenna used to image the 210m line from the shore to the ice (line [1], purple line on the map above). The ice/water boundary is hardly discernable. The lower image below shows the results of the 500MHz antenna imaging that same line, plus an additional 200m (line [1] purple plus line [2] green on the map above). There is a conspicuous boundary ~1m below the surface, which is effectively uniform. Both the depth and near-uniform shape are in agreement with the drilling data. Ground Penetrating Radar 1 – comparing 1,000MHz & 500MHz Herman, R. (2001) “An Introduction to Electrical Resistivity in Geophysics,” American Journal of Physics, Vol. 69, pp. 943-952. Kovacs Enterprises, https://kovacsicedrillingequipment.com/ Loke, M. H. (1996-2002). Tutorial : 2-D and 3-D electrical imaging survey. http://www.geoelectrical.com / Robinson, E. S., & Coruh, C. (1988). Basic exploration geophysics. New York: Wiley. DS18B20 datasheet: https ://datasheets.maximintegrated.com/en/ds/DS18B20.pdf MLX90614 datasheet: https:// www.sparkfun.com/datasheets/Sensors/Temperature/MLX90614_rev001.pdf References • This work funded in part by the Radford University Scholar Citizen Initiative (SCI) and the Office of Undergraduate Research and Scholarship (OURS). • Logistical support in Utqiagvik (né Barrow), Alaska provided by UIC Science. • Logistical support in Utqiagvik (né Barrow), Alaska provided by ARM Climate Research Facility. • The authors would like to thank the following for their contributions to the data acquisition: Tyler "Jamal" Bowman, D. Jake Clary, Jordan Lynn Eagle, Nolan McGrady, Sam Mogen, Nicholas Schrecongost, Rudolph Soltesz, Hans Voll, Abdullah Zulfiqar Acknowledgements C21A- 0662 1,000 MHz A visual correlation was observed between the surface temperature and the subsurface modeled from the “traditional” resistivity date. The image below shows the results of a resistivity line starting on the shore and proceeding onto the sea ice. The resistivity data were obtained with 4 receivers, yielding 4 data depths. Resistivity data were obtained every ∼ 0.4 along the survey line. High resistivity (ice) is indicated by the light blue while low resistivity (water) is the cark blue. Temperature data from the IR sensor were obtained every 88 along the same survey line. The first ∼ 35 of the line was on the shore with dirt/gravel observed under the ice/snow cover. This section showed the lowest temperatures by several ° . An abrupt change occurred when crossing onto the sea ice – it is likely that the sea ice was grounded until ~50, or ~15 beyond the “beach line.” These higher temperatures then generally follow the resistivity-estimated ice/water boundary for the remainder of the 200 survey line. The estimated location of that boundary is indicated by the red line (drawn in by hand). Note that the wide variation in the modeled resistivity data is an artifact of having too few vertical data depths. Future work should alleviate this issue. Note that in this image the highest elevation of the shore at the left of the image was ~1.2 above the ice surface. This elevation is not shown on the resistivity image below because of the large difference in horizontal vs. vertical distance scales. This simplification allows a better comparison of the resulting resistivity model to the surface temperature values. Comparing Surface Temperature and Subsurface Composition Via Resistivity -22.00 -21.50 -21.00 -20.50 -20.00 -19.50 -19.00 -18.50 -18.00 0.00 20.00 40.00 60.00 80.00 100.00 120.00 140.00 160.00 180.00 200.00 March 1, 2016 IR surface temperatures (C) Surface 4 per. Mov. Avg. (Surface) IR sensor in pvc pipe odometer wheel DS18B20  T sensors control & data logger unit position marker switch -22.0 -21.5 -21.0 -20.5 -20.0 -19.5 -19.0 -18.5 -18.0 0 50 100 150 200 Temp (°C) Distance (m) IR LOESS 6 per. Mov. Avg. (IR) The IR readings from line [1] (purple line) are plotted below with the measured data points (blue), a moving average of six (black), and a Local Regression (LOESS, red). The LOESS was made by linearizing three points on either side at a given x-location, and then interpolating a new value for that location. The average error (magnitude only) for the linearized points was 1.13%. The graph indicates the LOESS reacts to changes more quickly than the moving average, although we cannot speculate at this time as to the utility of this. Future work will include analysis using LOESS or a LOWESS (Locally Weighted Regression) methods. Further Sensor Sled IR Temperature Analysis [1] 200m line with 1 st ~35m on shore [3] 200m line, depths verified by drilling [2] 200m line “Expanding dipole-dipole” locations BARC Ilisagvik College One receiver was used with the transmitter of the capacitively coupled resistivity array to create an expanding dipole-dipole array. The dipoles were in the 2.5m configuration. The initial transmitter/receiver separation was 1.0m. Data were recorded at 1.0 second time intervals. Twenty data points were recorded at the initial 1.0m separation. After 20 seconds, the array was expanded by 10cm in either direction (this took ~5 sec) and the next 20 points were obtained. The transmitter/receiver separation was increased by these 20cm increments until reaching a final value of 10.4m. The entire process took ~25 minutes. We were able to discern when the separations were increased through the presence of a marked change in the mV/mA ratio recorded by the array. In addition, an identifiable “stable region” in the data revealed the ~20 data points with almost identical mV/mA values. These were averaged for each transmitter/receiver separation and plotted below. The data plots below use the “cumulative resistivity” method for vertical electrical soundings yield an approximately straight line for each layer in the subsurface. When one layer gives way to another, the slope of the line changes. Tracking the location of that slope change downward to the horizontal axis yields the depth to the boundary between subsurface layers. This method was applied at two locations on the sea ice. Both were along the line [2] on the map (the green line). One location was at the 210m mark (~175m offshore), while the other was at the 310m mark (~275m offshore). In both instances the depth to the bottom of the ice was measured with a drill and an ice thickness gauge [Kovacs]. In both cases the ice thickness was approximately 91cm. This is in agreement with the first slope break on the graphs below. A second slope break was observed on the graphs below. This break was unexpected and unfortunately observed after leaving Utqiagvik (né Barrow). However, while carefully measuring the ice thickness our group casually extended the ice thickness gauge to the ocean floor. One author (KM) informally noted that the depth to the ocean floor was roughly ~1.4m meters in both cases. While this was not a measurement it is consistent with the location of the second slope break in both graphs. There appear to be other slope changes in both graphs. However, without having measured these depths carefully, or obtaining seafloor samples, we cannot speculate about possible reasons for these variations. Expanding Dipole-Dipole Array & Ice Thickness 0 2,000 4,000 6,000 8,000 10,000 12,000 14,000 16,000 18,000 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0 Sr (Wm) effective depth z_eff(m) March 8, 2016; expanding dipole-dipole, 210m mark (sea ice) Sr vs. z_effective n>=0.60 n<=0.52 1 st slope break 2 nd slope break -100 -80 -60 -40 -20 0 0 20 40 60 80 100 120 140 160 180 200 ice depth (cm) distance (m) March 3, 2016: Measured (drill) ice depth of 200m line parallel to shore We used an ice drill and depth gauge [Kovacs Enterprises] to measure the ice thickness in order to validate the depths modeled using the geophysical data. We drilled at 5m intervals along the 200m-long line [3] (brown line on the map) approximately parallel to the shoreline. The data are on the graph to the right. This supported the GPR results of ice ~90cm thick with little variation. Ice Drill Results The 500MHz antenna – seen to be more effective at imaging the ice/water boundary – was used to survey line [4] (purple line on map) to compare with the surface temperature data from the IR sensor. This comparison to surface temperature was done with 2 sleds. The sleds were on either side of the GPR unit. They started line [4] within 5 horizontal meters of each other, but ended nearly 50m apart. The two sleds formed what could be described as a long, thin “V” shape with the point nearest the shoreline. Sled A (blue data points below) was further to the south and was pushed several meters further than sled C. Proceeding at an acute angle from the shoreline placed more of this line on grounded ice. There were a number of features that appeared to be related on the temperature and GPR images. • As with the resistivity image the grounded area showed a generally lower temperature. These near-shore readings also follow each other remarkably well with similar patterns of temperature increases and decreases. • The temperature readings noticeably rose and flattened out near the 760m mark. • The temperature readings showed a number of visually similar features, some of which are indicated with red arrows. • Another research group surveying the same area (with airborne lidar) asked us if our equipment had “…seen anything unusual at around [this 760m] distance from the shore.” We have communicated our data to them but we have not had time for further contact. Ground Penetrating Radar 2 – Comparison With Surface Temperature -27.00 -26.00 -25.00 -24.00 -23.00 -22.00 -21.00 0 100 200 300 400 500 600 700 800 900 T (C) distance (m) March 11, 2016 Sled A IR surface T readings Sled A IR 6 per. Mov. Avg. (Sled A IR) -27 -26 -25 -24 -23 -22 -21 0 100 200 300 400 500 600 700 800 900 T (C) distance (m) March 11, 2016 Sled C IR surface T readings Sled B IR 5 per. Mov. Avg. (Sled B IR) 500MHz GPR, 900m line [4] (blue line on map); ice/water boundary ~90cm below surface 0 2,000 4,000 6,000 8,000 10,000 12,000 14,000 16,000 18,000 20,000 22,000 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 Sr (Wm) effective depth z_eff(m) March 11, 2016; expanding dipole-dipole, 310m mark (sea ice) Sr v. z_effective n>=0.60 n<=0.52 1 st slope break 2 nd slope break 500 MHz ice/water boundary Note change in character of boundary in this area