Verification of Equipment
From GeoMod
Contents |
Chapter 2
Methods
Introduction: In order to determine the geomorphic effects of groundwater on the river system, it is first necessary to detect where the groundwater is discharging at a substantial enough rate to quantify. This will be initially done using traditional stream gauging techniques in which discharge measurements are made at various locations along the length of the river. Any gain or loss in discharge measured will be attributed to aquifer discharge or recharge. Once a gaining section of river is located that is conducive to study, a smaller area will be chosen for a finer scale analysis of discharge along the bed’s surface. Temperature has been used as a natural groundwater tracer since 1904 and will therefore be used in this study for its ease in use and detection. Water in the ground equilibrates with the temperature of the surrounding sediment. This ground temperature is the average annual temperature for a given area that exists below the area affected by diurnal temperature change and above any significant source of geothermal heating. Groundwater that makes its way to the surface retains the residual temperature of the subsurface. A device will be constructed to detect the river bed temperature and then the device will be verified by means of physical measurement of groundwater seepage. The latter will be done using seepage meters.
Since the thickness of the confining units that outcrop throughout Shelby County have variable thickness and outcrop location, it becomes necessary to first locate areas of groundwater discharge. The primary method that will be used utilizes the temperature difference in groundwater and surface waters. Since subsurface water equilibrates with the 62.7° F surrounding lithology determined by initial field investigation, groundwater maintains a constant ground temperature. The temperature of ground water near the Loosahatchie River was determined by lowering a Solonist Data logger into USGS well SH:W-23, 28 feet below the ground elevation. This well is immediately adjacent to the Loosahatchie River near Arlington, TN and is screened in the Memphis Sands Aquifer. Temperature was logged for a 24 hour period. No diurnal fluctuations were found. A plot of the logged temperature can be found in Figure 2.1.
During summer months, the temperature of surface waters in lakes and rivers averages well above this, and during winter months surface water temperature can be well below the 58° F ground water temperature. Gaining streams receive water from groundwater sources, therefore areas in the river that deviate in temperature from the surface water average towards the ground water average can be inferred as areas of groundwater discharge. Justification of this inference is described further in the chapter. For the purposes of this study, this deviation in temperature will be used as a proxy for groundwater discharge location.
Locating a Study Area
Locating a suitable study area will be done in two phases. The first phase will consist of using stream gauging techniques to find areas along the Loosahachie River that are gaining. This will be done using the wading method of stream discharge measurement utilizing a Swoffer velocimeter and a 100 foot roll measuring tape. This method has a base error of 5%. If the measured change in discharge is not outside of the 5% error range, a subsequent measure will be made using multiple measurements at each location to determine a more narrow range of error. The discharge value of the measuring point upstream will be subtracted from the current measuring point to determine the amount of water entering the system be means of the aquifer.
The second phase will consist of visually analyzing each gaining reach for factors that would make study difficult, such as large debris in the river, tributaries entering the river, municipal drainage pipes, and continued anthropogenic modification. The ideal study area would be substantially gaining and free of all aforementioned factors.
Stream gauging was performed on September 10th, 2006. A 1.8 mile section of river was measured for discharge between the Highway 70 bridge and the Collierville-Arlington Rd. bridge. Measurement locations are located in Figure 2.2.
One discharge reading was recorded for each measuring location. The discharge measurements were then plotted against distance to determine discharge rate. This plot can be found in Figure 2.3.
Discharge rates were measured on a larger scale on October 11th, 2006. Rates were recorded at the 195 bridge, Highway 70 bridge, Collierville-Arlinton Rd. bridge, and the Brunswick Rd. bridge. Discharge was measured at each location three times to get a better sense of the range in error. A plot of the discharge rates measured along with calculated error can be found in Figure 2.4.
Analysis: The discharge values recorded on 9/10/06 have a 5% error range associated with them since only a single discharge measurement was taken at each location. With this error range, it is possible that no gain or loss in the river occurred over the section of river measured. This prompted the larger scale measurement of the river on 10/11/06. Performing three discharge measurements at each location reduced the error to around 1-2%. This reduction in error makes possible the calculation of gain or loss to the river.
Discussion: Two river sections were found to be gaining; between 195 and Hwy 70 as well as between Hwy 70 and Collierville-Arlington Rd.. The First section is gaining at a rate approximately three times that of the second. The first section contains a drainage pipe from the Galloway city water treatment plant discharging an unsteady amount into the river. There is also a very large amount of fallen trees and debris in the river channel itself, obstructing flow and creating dams and pool sections. The second section has no major tributaries or municipal drainage pipes entering into it and is relatively free of debris. There is also easy access to bridges and a logging road that parallels the river. For these reasons, this second section between Hwy 70 and Collierville-Arlington Rd. will be used as the primary study area. Approximately ¼ mile downriver from the Hwy 70 bridge, the smaller 30x30 grid will be placed. It is sufficiently downstream to escape bridge effects and accessible for repeated weekly study.
Temperature probe Construction
Introduction: The probe needs to be able to acquire an accurate and consistent temperature at various locations in a river environment. These locations include but are not limited to river water surface, river bed surface, mid river depths, and sub-riverbed positions down to 30 cm. The probe needs to be water proof and corrosion resistant since the chemical makeup and red-ox potential of the water is not known. A data logging function will also be needed to record temperature data over extended periods of time ranging from 30 seconds to several days. Portability of the probe will be an important consideration since the majority of probe usage will be in remote areas in the field.
Meathods: Phase 1:
The initial design of the probe utilizes a ½ inch diameter stick of rigid electrical conduit cut to a length of 6 feet. A ½ inch threaded compression coupling was attached to one end of the conduit. A 6 inch diameter round stainless electrical box cover was then attached to the ½ inch compression fitting through the center punch out hole in the box cover. Two 3/8 inch holes where drilled in the box cover ¼ inch from the central punch out hole on opposite sides. The function of the box cover is to create a location along the pole that restricts penetration. In other words, the pole will be inserted into the riverbed until the bed comes into contact with the plate, which will restrict insertion. This creates the base to which the sensing and recoding equipment is attached.
A thermocouple probe is then inserted into each of the 3/8 inch holes drilled into the box cover. The thermocouple probes are type K probes with a ¼ inch stainless steal sheathing. Both probes are also grounded to the sheathing and sealed with a rubber handle with 4 foot extension cable. The grounded thermocouple gives a temperature response time of less than 2 seconds for type K thermocouples. One probe is 24 inches in length while the other probe is 36 inches in length. The probes have a +/- 0.5 degree Celsius bias and are accurate to 0.1 degrees Celsius. The probes are held in place with two pipe clamps a piece which are tightened around the handles of the thermocouples. This allows the thermocouples to be adjusted up and down the pole, adjusting the depth of penetration of each thermocouple. Clamping to the handle prevents the pole from direct grounding contact with the thermocouples. Two wire type K thermocouples are then attached to the conduit by means of pipe clamp at varying locations. The pipe clamps can then be moved up or down the pole to obtain measurements at different depths in the river water. Each thermocouple is insulated with a waterproof polyvinyl sheathing with an exposed junction. The exposed junction allows rapid temperature response of less then 1 second.
An Omega model HH309 four channel data logger is then attached to the top of the conduit. The HH309 can record temperature in Celsius or Fahrenheit on four separate channels. It has a resolution of 0.1 degrees Celsius. It can store up to 16,000 records per channel and has an NIST traceable Certificate of Calibration. It can log temperatures form as frequently as 1 measurement a second up to one measurement every 2 hours. Each of the four thermocouple leads is then numbered and plugged into a port on the HH309.
A measuring tape is then taped to the conduit with the base, 0 cm, end of the tape at the box cover. The other end is taped to the top end of the conduit. This creates a marked reference along the length of the conduit for obtaining depth of water data.
Phase 2:
The two thermocouples with the stainless steel sheathing are removed. This is due to issues encountered when the thermocouples are inserted into the sediment. Readings became very sporadic and range in reading differed by several hundred degrees, an impossibility for natural conditions. This is likely due to electrical interference induced by the iron in the sediment interacting with the stainless steel sheathing to which the thermocouples are grounded. Two additional wire type thermocouples are added to the probe in lue of the stainless steel sheathed thermocouples. These are attached to a 24 inch length of ½ inch electrical conduit with a ½ inch threaded compression coupling on one end. The exposed junctions of the wire thermocouples are covered with two layers of gauze cloth. The cloth is then held in place with Duct tape. The thermocouple junctions are located at 5 centimeters and 20 centimeters form the compression coupling. A ½ inch threaded coupling is then used to attach this new conduit piece to the existing probe assembly. The two new wire type K thermocouples are then plugged into the HH309 data logger.
Phase 3:
The 24 inch length of ½ inch conduit is removed from the probe assembly. It was deemed unnecessary after comparing various depth data with bed surface data. A sufficient correlation was made to validate using bed temperature data which can be acquired much faster and much more easily than sub bed temperatures. The two wire type K thermocouples were removed from the 24 inch length of conduit. One was attached to the underside of the box cover to measure the temperature of the bed surface. The other was attached to the top of the box cover to measure water temperature immediately above the river bed. These thermocouples where then plugged into the HH309 Data logger.
Results: The final temperature probe was able to record the river bed surface temperature accurately down to 0.1 degrees Celsius. It had a response time of around 1.5 seconds. Temperatures measured on the bottom of the contact plate were generally within 0.1 degrees Celsius of the temperatures recorded on the top of the contact plate which was in contact with near bottom river water. Depth was measured to an accuracy of about 1 cm. This is due to slight river water surface fluctuations.
Discussion: The probe works well for measuring depth of water and temperature of river bed. While it can detect slight variations in bed temperature, there is no verification that these variations are indicative of ground water discharge rate. Seepage meters will be used to determine if the temperature variations correlate to physical discharge rate measurements made with the seepage meters.
Seep tank Construction and Usage:
Introduction:
The goal of this project is to determine rates of vertical discharge on a much finer scale, creating the need for smaller seep tanks with a smaller cross sectional area. These smaller tanks could then be used to show any correlation that may exist with the point temperatures obtained from the riverbed. If a correlation is found, temperatures may offer a proxy to seepage meter tanks. This is beneficial since seep tanks require a substantial amount of time compared to the nearly instantaneous temperature measurement of the riverbed.
A seep tank is simply a container with an open end with is inserted into sediment to measure the amount of fluid moving vertically. This is accomplished by creating a hole in the top of the tank and attaching a reservoir of some sort, usually a bag, to catch effluent water. The amount caught is then divided by the time the tank was in position and divided by the cross sectional area of the tank to obtain a rate of fluid movement vertically. Seepage tanks are usually used to determine rates of groundwater being discharged into a stream, however, it is also possible to use seepage tanks to determine rate of recharge. This is done by adding a known volume of water into the reservoir bar before insertion of the tank into the sediment. The amount of water in the reservoir is then measured after a known amount of time has passed and the difference in volume divided by the time that the tank was in the sediment is the measured recharge rate. Historically seep tanks have been constructed from 50 gallon drums that have been cut in half, creating two short cylinders. A hole is the n drilled in the top of each half and a plastic bag is attached to the hole by means of a pipe fitting. Size of the fitting is irrelevant since discharge rates are small enough to negate flow impedance by friction in the pipe fitting. These tanks are then inserted to a maximum penetration possible. Five gallon buckets are a popular alternative to the 50 gallon drum for their portability and less expensive parts list.
Belanger and Montgomery designed a study in 1992 whereby they were able to test the accuracy of seepage meters in a controlled laboratory environment. A large circular tank, 3 meters in diameter, was filled with layers of gravel and sand to simulate a uniform riverbed. Seepage meters made from 50 gallon drums were then inserted into the model riverbed. Water was then forced up through the bed material and measured using the seepage meters. Error caused by friction at seepage meter walls was found to be negligible. While the rate for the tank was constant and the bed material was installed as informally as possible, variations in seepage rate were found throughout the tank. Tank discharge rates were found to be consistent at each seepage meter location, however, they were found to be more accurate if averaged in pairs. Even in a controlled environment, seepage rates were found to very widely in close proximity from seepage meter to seepage meter.
The seep tanks used in for this project are composed of three main components; 1)hexagonal plastic container with a cross sectional area of 146.1 Cm2, 2) 90 degree half inch compression fitting, 3) 1 gallon plastic freezer bag. The main tank is comprised of a plastic container in the form of a hexagonal prism. The end of the container which contained the lid was removed to form a hexagonal prism with an open end. It is 22 cm in height and 13.5 cm wide in cross section. Each face of the prism is 7.5 cm wide (See figure 2.6). A ½ inch diameter hole was drilled into the closed end of the container 1 inch from edge near a corner of the hexagonal top. A ½ inch, 90 degree compression fitting for electrical flex conduit was then attached to the tank by means of the hole drilled in the closed end. The standard threaded end of the fitting is inserted into the tank and held in place with a ½ inch electrical conduit nut. A plastic washer was placed between the fitting and the tank to ensure a tight fit and water proof seal. The fitting was positioned facing away from the center of the tank. A 1 gallon freezer bag is then attached to the tank by means of the compression fitting.
After some question into the reliability of the compression fitting was raised, the tank’s compression fitting used for attaching the collection bag was replaced. A ½ inch street 90 was attached to the seep tank using the same hole the compression fitting was previously attached by. A ½ inch close fitting with a ½ inch electrical nut was set from the inside of the tank, up through the hole and attached to the street 90. The bag was then attached to a separate ½ inch sprinkler male coupling by means of rubber band. The coupling can then be tightened onto the street 90 by hand for a secure and leak-free connection.
The tank is used by inserting the tank into the sediment with the fitting facing downstream. The collection bag is not attached until after the tank has been fully inserted. This allows river water to leave the tank while being inserted. Sediment is therefore not displaced and conditions of the sediment remain intact. The bag is then attached to the compression fitting and a stopwatch is used to record the time that the bag is attached to the tank. After a certain amount of time has passed, the bag is clamped off near the fitting and the stopwatch time is stopped. The bag is then removed from the tank along with the water collected. This water is then measured using a 100 mL graduated cylinder which has an accuracy of +/- 1 mL. This means that for every 100 mL of water collected, there is a 1 mL error associated with it. The amount of water is then divided by the time in seconds recorded by the stopwatch and then divided by the tank cross section in meters to obtain rate of discharge in mL/sec/m^2.
Methods: Tank Verification To verify that the tanks are functioning properly and accurately measuring discharge, a series of experiments were designed. The first experiment was to test that the tanks are accurate and obtaining similar rates at the same location. The second experiment tests the correlation between seepage rate and temperatures measured with the temperature probe.
Experiment 1: Method: Three locations are chosen in a predefined study area. Each location comprises a 0.5 meter diameter region of riverbed. Each area is marked with an orange marker stake to make recognition and recovery of seep tanks easier. Two seep tanks are then inserted at each of the three locations. Discharge rates are then measured by all seep tanks. Leaving the tanks inserted, rates are measured a total of four times for each tank. The tanks are clustered in pairs to verify that tanks at the same location are recording the same discharge. The repetition of measurement four times verifies that the amount measured by the tanks is consistent.
Results: Data acquired for experiment 1 is located in Figure 2.7.
Analysis:
Position of the tank utilizes the predefined coordinate system used in the temperature probe studies. Positions and data can therefore be overlain on existing maps. Three locations where chosen at random: (-20, 20), (10, 12), and (-4, 4). Seepage rates by location were found to very slightly between tanks. When the average of the two tanks was calculated by round, the average per location appears fairly consistent. The standard deviation for locations is as follows: (-20, 20) was 0.0250, (10, 12) was 0.0241, and (-4, 4) was 0.01616. This yields a percent error of 12.6% for (-20, 20), 14.4% for (10, 12), and 38.4% for (-4, 4). The ranking of each site remains intact for each round of measurements.
Discussion:
While tanks did not always measure near the other tank installed in their cluster, the averages remain consistent. In each cluster, a particular tank may be higher than its counterpart in one round and then be lower in the next, but the average is still preserved. This may be due to small fluctuations in the hydraulic gradient beneath the river, change in pressure due to migrating sand forms, or perhaps caused by the action of walking around near the cluster location before and after bag installation. A video camera positioned at various positions in the river recorded large sand ripples, 15 to 20 cm in height, migrating as much as one meter downstream in fifteen minutes. Fixed reference stakes placed at the head of dune features made possible the recording of dune migration rates as high as one meter in 8 hours. Removing or adding 20 cm of sediment along the river bed surface could have a significant impact on the flow dynamics immediately below the bed surface. It also indicates that any effects of walking which disturb the surface sediment will be quickly diminished by migrating sand. The speed of migrations also allows for an average state to be found much more quickly. These results indicate that installation of multiple tanks per location and repeated measurements will result in accurate seepage readings.
Expirement 2:
Method:
The three seep tank positions used in experiment 1 will also be used in experiment 2. The temperature of the riverbed will be taken at each cluster location. Temperatures and position will then be entered into Surfer 7.0 for graphic analysis. Discharge rates acquired by seep tanks and positions will also be added into Surfer 7.0. Using the Kriging method of interpolation, profiles will be generated for temperature and for discharge and compared. Numeric analysis will also be done to the raw data to determine the degree of correlation between temperature and discharge rate.
Results: Data entered into Surfer 7.0 created Figure 2.8 and Figure 2.9.
It is possible to see form the two generated profiles in Figure 2.8 and Figure 2.9 that an apparent close correlation exists between the temperature of the riverbed and the discharge rate.
The calculated correlation coefficient r for the three clusters over the 4 rounds is -0.98928. The regression can be seen in Figure 2.6. This is a nearly perfect inverse correlation between temperature and discharge rate. We can therefore validate the use of temperature as a proxy for discharge rate. The data suggests that use of temperature may be accurate enough to determine overall seepage rate for a given location. The original intention of Experiment 2 was to validate the rank correlation of temperature to the rank of discharge, but such a high degree of correlation suggests temperature may be used to determine actual discharge rate, not simply rank. Further studies may produce a means of using macroscopic temperature readings acquired by aircraft or satellite to determine discharge rate in certain river systems.
Further Investigations into Experiment 2: Methods: To further validate that seepage rate and bed surface temperature can be correlated, the experiment was conducted again several weeks later. River water temperature was 1.5 degrees Celsius cooler and flow rate was comparable to the first experiment. To obtain a more robust data set, six measurement locations were chosen. In this round, the seepage tanks were modified slightly to ensure that no water was leaking from the tank through the collection bag fitting. The compression fitting used to attach the seepage bags was replaced with a screw on pipe fitting to ensure that the bag is attached securely.
Tanks will again be put into location in pairs to preserve an average for each location. Since only six seepage tanks were constructed, the measurements will be done in two sets. Three pairs of tanks for three locations will be used for the first set. The tanks will then be moved to three new locations for the second set. Locations of tank pairs will be determined randomly before any tank insertion and marked with a red marking stake.
Temperature of the surface of the bed will be acquired for each of the six locations before insertion of seep tanks. It will be recorded again after the first set is completed. Sub Bed temperature will be recorded for each of the six locations after the second set is complete.
Results: Data for experiment 2 is located in Figure 2.10.
Discharge rates and temperatures of river bed at seep tank locations were entered into Surfer 7.0. Using the Kriging method of interpolation, a topographic profile was created for the discharge rate and temperature difference. Since only 6 tanks were available and they were to be installed in pairs, three pairs of tanks were installed in 3 locations, temperatures were recorded before and after insertion, and then tanks were moved in pairs to 3 separate locations. Since the flow effect on temperature can be affected by tank insertion, only temperature recorded for each position after tank insertion were used. Since water temperature increases through the course of the day, the absolute temperature of the first three locations was lower than the second three locations. In order to preserve the correlation of the six tanks, temperatures recorded for processing were adjusted by subtracting the temperatures measured at the seepage tanks from the temperature of the river water at the time the measurement was taken. Figure 2.11 is the temperature difference form datum determined by kriging. Figure 2.12 is the discharge rate determined by kriging.
Analysis: This round of seep tank testing yielded an average of 15.8 percent error. Percent errors ranged from 11% to 21%. These were calculated by analyzing each sample point discharge with the average discharge rate. The increased range in error was expected since the tanks were installed in two separate rounds and the increased number of positions allow for more variation in bed materials encountered. Overall, the rankings of the seep tank discharges correlated very well with the temperature measurements made after tank insertion. The second experiment yielded a correlation value of r = -.81. When values from both sampling days were combined the regression value became -0.79. There is a definite variation in slope of regression that is dependant on the temperature of the river water when the measurements are recorded.
Discussion: The second experiment using twice as many tank locations as experiment 1 yielded very similar results. This makes the results accurate and reproducible. In both experiments, the seepage rates were nearly perfected inversely correlated to temperature measured on the riverbed surface. The values obtained after altering the tanks were very similar to those recorded before tank alteration, indicating that the bag seal in experiment 1 was sufficient and values obtained for experiment 1 are accurate. While the compression fitting has been shown effective, the threaded pipe connection is much simpler to assemble and easier to attach to the tank when underwater. With a high degree of reliability in the seepage meter experiments, it is valid to state that temperature of river bed surface can be used as a proxy for groundwater discharge rate. The variation in regression slope that is dependant on river temperature can be eliminated by sampling bed temperature below the effect of diurnal heating and cooling. Response and stabilization times for these deeper measurements can be on the order of several minutes, making the rapid reproduction of multiple measurements not feasible. Since this study is looking for a relative rate of discharge along the river bed surface, the less time consuming bed surface temperatures will be used.
