Characterization of the thermal wake of a weather balloon during ascent

Research Group: Julia Abbott, Matthew Harlan, Mateo Reynoso, and Hannah Tompkins

Launch: Whitworth Spring 2017

''We chose to study the shape of the thermal wake because it is a relatively new area of research for this near space class. Our intention was to examine the reasons for the temperature behavior due to the balloon during ascent into the stratosphere. However, during the launch, our pod only collected twenty minutes worth of data so we were unable to analyze temperature trends above the troposphere. From the data we did collect, we were able to make observations about the nature of the digital DS18B20 temperature sensors that we used. From those observations, we made some suggestions to future students about calibration and handling of these sensors.''

Background
There are many layers to the atmosphere. Of these, those that we are primarily interested in are the troposphere, which is the lowest, and the stratosphere, which comes right after it. Both sections have an upper layer: for the troposphere it is called the tropopause. Likewise, the upper layer of the stratosphere is called the stratopause. When ascending through the atmosphere, the temperature begins to decrease to approximately -40 degrees Celsius until hitting the stratosphere. From there, the temperature begins to increase to about 0 degrees Celsius until one reaches the stratopause. The temperature increases due to the absorption of ultraviolet rays by the ozone in the stratosphere. In past experiments in near space, results for temperature measurement have been much warmer from what is to be expected according to scientific journals. For example, in a section of the stratosphere where the temperature is expected to be about -50 degrees Celsius, the temperature was instead measured to be 10 degrees Celsius. These results began to make sense in the context of a thermal wake that the balloon generates when it is warmed up by the sun. Thermal wake occurs when the sun heats the surface of the helium balloon that takes all the experiments into the stratosphere. The heated air siphons off the surface of the balloon as it rises and ripples outward from the string of experiments like the water wake from a speed boat as it goes through the water. This results in a bubble of warmer air fanning out from the tail. According to Erik Agrimson and Kaye Smith in their research, thermal wakes extend at least twenty feet from the base of the balloon. In order to better define the physical boundaries of the thermal wake, a pod will be sent up with two arrays of sensors. This will allow us to analyze the results of any future experiments involving thermal behavior with a greater understanding of the context.

Mechanical Design
To measure the edge of the envelope of warmer air around the experiments, we attached a string of DS1B820 temperature sensors along two carbon fiber rods that protrude about eight feet from two lateral sides of our pod. These sensors are spaced roughly four to six inches apart in order to determine where the warmer air meets the wall of the characteristically cold air of the stratosphere. Our carbon fiber rods are three millimeters in diameter and are therefore prone to bending, which can distort our gathering of data. To account for the rods flexing, we attached a string to the end of each eight-foot rod and anchored the other end of the string to the command module above our pod. Because each rod is about eight feet long, we had to assemble the pod with the arms on the launch site. We prepared to make this assembly as quick and reliable as possible by zip-tying foam-lined plates of metal inside the pod on either side. On site, the rods were inserted in holes in the side of the pod and then clamped down with the metal plates. To measure the properties of the thermal wake below the balloon, and its string cargo, we attached our pod last below everyone else's. To the bottom of our pod, we attached another long string of DS18B20 sensors. Unlike with the sensors on the arms of the pod, the sensors on this string are about one meter apart because we expected the very end of the wake to extend far below the end of the last pod. On the first arm(arm number 1), we soldered 16 DS18B20 sensors together. The second arm (arm number two) has 18 sensors to match the total length of arm two, and the tail has 11 sensors soldered together. We secured the wires of the temperature sensors to the arms with zip ties and the wires entered the pod in a small hole above the hole where the carbon fiber rod enters the pod. Inside the pod, the wires from both arms and the tail plugged into the micro-controller. To prevent the micro-controller from bouncing around inside the pod during flight, we used screws and washers to secure it to a board which was attached to the bottom of the pod with zip ties threaded through the Styrofoam of the pod. The battery for the micro-controller was secured similarly.

Electrical Design
The wires from both arms and the tail connect to the microcontroller via power, signal, and ground wires—one for each arm and the tail. The sensors are soldered in parallel to each other and then plugged into the breadboard. Digital Inputs 45 DS18B20 temperature sensors Circuit Total current: 142.2 mA

Battery Life: 4.9 hours. We found this by dividing the 700mA hours value given by the manufacturing data sheet by the total current load of our circuit of mbed and 45 DS18B20 sensors. We found the total current load of 142.2 mA with an ammeter

Code
The code for this project relied on several dependencies which are as follows: a modified DS1820 library, a library to mount the SD card successfully, and an extended timer. It also includes the standard mbed library. The code relies mainly on loops to execute the majority of its functions. Immediately after startup, the program attempts to connect with the SD card. If it is unsuccessful, the program exits with a code of -1. If the SD card is successfully mounted, then the program proceeds to start the timer. The program then creates three different arrays of DS1820 objects, which hold all of the sensors on a given arm or tail. Three separate files are also created which will be situated on the SD card and used to store the data from all of the sensors. After these actions have been taken, the program enters a while loop for each of the three strings of sensors, with the exit condition being to assign all of the unassigned probes to its arm's array. The loop is also set to break if the maximum number of probes has been met for any given array. In this loop, all of the sensors on the given arm are assigned to its personal array. These loops execute in consecutive order, rather than at the same time. Another of series of while loops execute after these tasks have been completed, and print out each of the sensors' unique identification number. Finally, the code enters an infinite while loop that continues reading temperature readings from each of the sensors on each of the arms.

Program source code.

Validation and Calibration
Our original plan for calibration of the sensors involved placing five randomly selected DS18B20 sensors in both hot and cold environments and then comparing the resulting measurements to those from three reliable temperature sensors. This group of three included: two voltage thermometers (whose equipment numbers are 002116 and 001427) and one mercury thermometer. The measurements from the five DS18B20 sensors were consistent with each other despite differing significantly from the measurements of the calibrated sensors, so we felt comfortable using the measurements from our five sensors to make a broad calibration graph for all forty-five of our DS18B20 sensors. To account for these discrepancies between these two methods of measurements, we found equation for the line best fit of the graph for the calibrated group of three versus the DS18B20 sensor values. The equation is as follows: y = 1.25x-4.49. To convert the DS18B20 temperature measurements to calibrated measurements, we then inputed the DS18B20 values in for x to get the corresponding y-values as the calibrated temperatures. In short, we had a single scaling factor that we applied to all of our DS18B20 sensor measurements.

After examining the temperature data from these sensors on a graph of temperature versus position from the pod, we realized that our previous method of calibration was too broad to apply to all the temperature sensor measurements. It became clear that each sensor at the same moment in time did not collect the same temperature values. The five sensors that we had randomly selected for our first calibration gave relatively similar temperatures but, in retrospect, that was not a large enough sample size to gain a sense of how variable the measuring behaviors of all forty-five of the sensors were. With these new observations in mind, a more accurate calibration technique would place all of our DS18B20 sensors in hot and cold environment along with the three reliable sensors and made a calibration graph similar to our first one but for each DS18B20 sensor. In this way, each sensor would have their own scaling factor for the temperature measurements. If we had used this more specific calibration method, we expect that our data graph of temperature versus position from one point in time would display much more consistent temperature values.



Data and Analysis
Temperature Analysis of Arm 1

Arm 1 is one of the lateral arms that protrudes for the side of the pod. There are sixteen DS18B20 sensors along Arm 1 and in this graph, we chose five different periods of time from our data to plot the temperatures from each of the sixteen sensors. This allowed us to see the temperature spatially from the pod out to the tip of Arm 1. This graph format would have been ideal for measuring the thermal wake, which was our intention. As mentioned in the calibration section of the wiki, the temperatures at one point in time should not vary between the sensors. So, while we were unable to analyze any temperatures from the stratosphere, with the twenty minutes of data from the launch field, we were able to adjust our calibration techniques to better represent the varying nature of temperature measurement among the DS18B20 sensors. Also, from the graphs of temperature versus position for each arm and the tail, we noticed that as time increased, the temperature also increased significantly--even in the twenty minutes that our pod was collecting data. We assumed that this happened because the sensors are made of black plastic and heat up quickly in direct sunlight. With this in mind, any future students who would like to use DS28B20 sensors could prevent this heating by painting them white. This would allow them to interpret any increase in temperature to something other than radiation, like thermal wake!

Temperature Analysis of Arm 2

For Arm 2, which has eighteen sensors, the temperature trends over time were fairly similar to those of Arm 1, which reinforced our assumptions about the behavior of the sensors when they sit in direct sunlight in an open field. Also, the relatively synchronized nature of the temperature curves over time suggest that the sensors along the arm were exposed to the same temperature in the same instant of time, the discrepancies are instead attributed to the lack of specialized calibration.

Temperature Analysis of Tail

For the tail, the temperature curves aren't as synchronized with each other as with Arm 1 and Arm 2. We think this is the case because the tail wasn't attached to a rigid carbon fiber rod like Arm 1 and Arm 2. Especially toward positions of 400 to 100 centimeters from the bottom of the pod, the curves from the five moments in time cross over each other--even though the sensors should have recorded the same temperature for a single instant in time. This likely occurs because the tail sensors were laid out on the ground in a less organized way than the sensors for Arm 1 and Arm 2, with some the sensors directed into the cooler grass and some pointed straight into the sun. Looking at the temperature values for all five points in time but between zero and 200 centimeters from the pod--the most reliable position on the tail because it is secured so closely to the stationary pod--we see the same warming trend over time that took place with the sensors on Arm 1 and Arm 2.

Conclusion
In graphs of temperature versus position along each arm and the tail, we expected to see warmer temperature measurements from the sensors closest to the pod, and measurements of about –30 degrees Celsius from the sensors at the farthest ends of the arms and tail. Results like these would indicate that there indeed is some sort of wake that encompasses the area below the helium balloon. However, after inspecting our data, we discovered that after about one thousand seconds—roughly twenty minutes—the code stopped recording temperatures. We suspect that this happened because we didn't secure the screw terminals into the bread board very reliably.

Our method of winding electrical tape around the bread board after inserting the screw terminals was not sturdy enough to prevent the wires from coming out once the pod was subjected to wind turbulence. While we did test the reliability of this method by pulling at the wires before the launch, we should have accounted for the harmful effects of gradual shaking and vibration of the bread board over time.

Within twenty minutes of starting our code during launch day, our pod didn't quite get the chance to collect any temperature from the troposphere or stratosphere. This means that the majority of the collected temperature values were from the empty sports field that served as our launching pad. While we were unable to analyze thermal wake with this data, we noticed some trends from the sensors that allowed us to draw some conclusions about the temperature-measuring behavior of DS18B20 sensors.

Even though we made some useful observations about the properties of DS18B20 sensors, here's what we would do differently in our electrical design in order to successfully research what we had originally wanted to study: thermal wake. The main issue was with securing the strings of sensors via the screw terminals to the bread board. The prolonged shaking of the arms placed strain on these screw terminals inside the pod which likely resulted in the wires coming loose from the bread board. To prevent this tension, we would increase the length of these screw terminal wires inside the box. In this way, the wires would stay firmly inside the bread board in spite of the shaking. To increase the organization and sturdiness of the bread board, we would also solder the wires on our circuit.

Another change we would make to our design concerns the code. Much of our troubleshooting during the testing phase of this experiment was prolonged because we had difficulty identifying a problem if the data wasn't being collected correctly. In order to make diagnosing—and by extension, fixing—any errors in code or assembly more efficient, we would set up a part in the code to flash an LED light if something specific went wrong. By setting up multiple alerts for different scenarios of how the code or wiring could fail, it would be easier to pinpoint the issue and fix it.

References/Works Cited
Erik Agrimson, James Flaten, Kaye Smith, Mara Blish, Rachel Newman, Juliana White, Maggie Singerhouse, Emily Anderson, Spencer McDonald, Christopher Gosch, and Alex Prat, Continued Exploration of the Thermal Wake Below Ascending High-Altitude Balloons, AHAC, 2014.

http://www.srh.noaa.gov/jetstream/atmos/layers.html

http://www.ces.fau.edu/nasa/module-2/atmosphere/earth.php (for layer atmosphere pic)

https://earthobservatory.nasa.gov/IOTD/view.php?id=7839

http://www.albany.edu/faculty/rgk/atm101/structur.htm