Weather Forecasting
This topic submitted by Brad Kalchek. Justin Hamill, Louie Banks, Andy Dooley (Kalchemb@muohio.edu) at 5:38 pm on 12/9/99. Additions were last made on Thursday, January 4, 2001. Section: Cummins
Student Generated Lab Instruction Packet
Weather Forecasting
Our problem is the weather. Weather is an ever-changing but consistent natural occurrence. Folklore says that weather brings magical things to those who are worthy of them. Weather was once worshipped as the deeds of the Gods by many pagan religions. When mere mortals wronged the Gods high on Mount Olympus, the Gods responded with lightning and rain and floods and various other events that would always have a horrid effect on the lifestyles of the Greeks. Native Americans also worshipped the weather because it was the bringer of rain and moisture, and that kept their crops alive. In fact, we are all very much affected by weather on a daily basis. What we hope to do in our lab is to make clear the patterns that are always present in our ever-changing skies. We will attempt to use minimal instruments to predict the weather on a daily basis. Our question is: Can you predict the weather accurately? And if so, how accurately? Our hypothetical answer being that, every aspect of our outside world, such as: wind direction/speed, barometric pressure, temperature, humidity, cloud formations, and cloud movement are all tied together and can be used wholly to predict the weather fairly accurately. We decided on our project through many hours of deliberation and decided that we wanted to do a very observatory project. Many of our other ideas were maybe too confined to a simple trial and error theme and we wanted something that would give us seemingly endless amounts of information to calculate. After we had decided on the subject matter of our project (weather), narrowing that down to a single question was easy. Finally we decided to try our best at predicting the weather through a series of daily observations and predictions. Our plan was to use the concrete information that we get from all of our instruments and to compare that to the information that we predicted and see just how close we actually get to predicting the weather accurately. This routine became quite interesting because the information was always changing so we had to cater to that change by changing our predictions. It was also interesting to see if our predictions had come close to what really happens on any given day. After all the hard work of observation and calculation, we had fun trying to keep up with the crazy Ohio weather.
Materials and Methods
Predicting the weather can become a very frustrating thing to attempt without the proper experimental design. Our project started completely off of one observation. The project started on a Tuesday, hence our datasheet did also. That Tuesday morning, we observed the aesthetic weather for the day. By aesthetic I mean what we could see outside, without having to use any weather instruments. So we observed the weather and then added our predictions for that following Wednesday to our datasheet. Then the real fun began. On Wednesday, our instrumental observations started. Four times a day, we would observe every aspect of our weather that was listed on our data sheet. After we had done this we could calculate the average temperature, humidity, wind direction, and so on. To make sure our data was sound, we checked it periodically with real-time weather radar for the Ohio area.
We ran into a problem however, observing the many types of cloud formations. There was really no way for us to “average” the many types of clouds that were present in the sky every day. There was also no way for us to statistically check the accuracy of our predictions on the subject of cloud formations and whether or not it rained. So, what we did was to design a point system for each type of cloud, which essentially just replaced the cloud type with a number, so a calculation and t-test could eventually be performed.
Materials
We tried to use minimal materials to perform our tests because the availability of them to the whole class would be limited, so we tried to stay as simple as possible. For the temperature calculations, we used a Celsius scale thermometer and took temperatures at various times of the day (morning [8:00], noon, afternoon [3:00], and sunset [7:00]). After the temperatures were recorded onto a notebook, an average temperature was taken and that became the temperature for the day. Humidity was calculated using a device called a sling psychrometer and a table that helped in calculating the percent humidity. This device measures the heat that is needed to evaporate water and can measure the moisture in the air. The humidity was also studied at the four intervals throughout the day, and an average was reached which became that day’s percent humidity. While studying the cloud formations, we used a web page that listed and described the main types of clouds. We used the page as a cloud map to identify the clouds that we were looking at on a daily basis. Wind direction was observed more practically, also. For wind direction, we periodically observed a real-time weather map on weather.com to see which way clouds in the atmosphere were moving. We also studied a surface wind map that outlined wind direction and speed.
For the most part, we feel that our experimental design is statistically sound. The fact that we used average temperatures and percent humidity calculated only four times daily, will possibly pose a problem in the accuracy of the observations but will not account for any real significant difference in the comparison between our predictions and our actual observations. This is true because, even thought the weather maps update and take measurements many more times a day, we still use the average of them too, so no real difference should be expected. The point system for cloud types and overcastting may also be awkward too, but should prove to be helpful in the t-testing of our predictions to our observations. We also received the consent of Professor Cummins to use this point system; he too felt that it would benefit the statistical testing. The consistency of our testing may also prove to be a offsetting factor in out testing. Getting a reading at exactly the same time of the day and dealing with classes can be quite difficult, especially with weather. To accurately predict the weather, you need an almost constant stream of updated information so as to observe the many changes that weather will undergo in any give period of time. We, however, only had minimal observations to work from, which will account for most of the error in our studies. With this in mind, the class’s observations and predictions will also be erroneous for the same reason that ours were. We do urge the class to be as consistent as possible in their observations to insure a more accurate reading in the final product.
The class, for the most part, will be asked to use our method of observing and recording and finding patterns in the weather to be able to, in the end, predict weather somewhat accurately. Sample data sheets will be handed out with each lab packet for the students to fill out over a week’s time. Students will also be provided with a list of Web Pages that were instrumental in our observations and will be provided with a cloud chart. Using StatView, this datasheet can be translated to the program; the students will conduct a t-test on each section of the datasheet. This must be done separately so as not to mix different types of information. The t-test will compare the accuracy of the predicted weather patterns to the actual observed weather patterns. This test will show whether each test was individually accurate.
Results
Upon completion of the lab certain conclusions could be drawn using the evidence collected. Through the usage of multiple t-tests between the predictions that we made and the actual weather information that we collected and the plotting of comparative graphs, we came to the conclusion that our hypothesis was correct. We hypothesized that weather can be accurately predicted through the use of limited tools and that all weather phenomena are in some way related and all had effect on one another. This is true because, after t testing, our p-values for each test were greater than .0005. Since our t-test results for each predicted vs. actual test had p-values that were greater than .0005, we could accept the null hypothesis and say that there is no significant difference between what we predicted and what actually occurred. Other information that we discovered from our research was that average temperature and average relative humidity are inversely related; as temperature goes up, relative humidity goes down which could be due to the fact that when humidity is high, the sky is cloudy which brings temperatures down. And when humidity is low, skies are usually clear, which leads to a higher average temperature. Average humidity and average pressure, however, and directly related; as humidity goes down, so does pressure. This is due to the fact that when pressure drops, it usually rains or clouds form, which in turn lowers the relative humidity because moisture is taken from the air and condensed. Precipitation and humidity are also related directly while using our point system for cloud type. When the cloud type hits point 5 (cumulo-nimbus), it is about to rain. High relative humidity is also a sign that it is about to rain because that means that the air has reached its saturation level and cannot hold anymore moisture so it condenses into clouds and eventually the clouds become saturated and it rains. This is also true as a clear day leads to a lower average relative humidity. Using the same point system for cloud type, pressure and precipitation are inversely related. When the cloud type hits point 5, which leads to rain, the pressure drops to a low level. Low pressure system is also related to rain as the moisture leaves the air and vapor pressure disappears from the air, thus making the air lighter (lowering pressure). Finally, temperature and pressure are inversely related. As temperature rises the pressure starts to drop. This is true because as warm fronts move through, the temperature rises but the pressure drops to meet the low-pressure systems that follow the warm fronts and bring rain. Upon completion and extrapolation of the cloud aspect of our lab, specific trends were discovered between cloud type, wind direction; cloud cover, and their participation within weather forecasting. WE were able to discover these trends by using two resources: one was our observation, which we compared in graph format, and the other was a meteorology web site, which we used to check our results. In order to successfully compare the cloud formation, and cloud type data we were forced to invent a number point system. This point system is what allowed us to graph and interpret our data. The number system was applied as follows: zero points were given on a clear day, one point was given on a day with cirrus cloud formations, two points were given to stratus cloud types, three points were given to nimbo stratus clouds, four points were awarded to cumulus cloud formations, and finally five points were given to cloud formations which appeared in the cumulo nimbus range. Cloud cover was analyzed in much the same fashion. Clear clouds were given a one, scattered clouds were given a two, a three was given to partly cloudy days, a four was awarded to cloudy days, and a five was given to days with heavy cloud cover.
Our analyzation and interpretation led us to discover the following results. Cirrus clouds, characterized by high ice crystals pumped high in the sky by a vortex, generally brought good weather. When compared to our wind direction data we discovered that this only occurred when the wind came from the West-Northwest to the North. Precipitation was likely within 20 – 30 hours if the same cloud formation was spotted along with a Northeast to south wind direction. Cirrostratus clouds brought precipitation within 15 - 20 hours if the wind came from the Northeast to the south or even sooner if the wind flowed from the Southeast to the south. Other wind directions seemed only to bring an overcast sky.
Cirrocumulus clouds are characterized with slightly worse weather conditions. If winds flow from the Northeast to the south precipitation was likely within 15 hours. If discovered during early morning, summer weather conditions, thunderstorms are likely. Altostratus cloud formations bring precipitation within the following ten hours if winds are steady from the Northeast to the South. Other wind directions result in overcast skies. Alto cumulus cloud formations, mixtures of water and ice, appear as a blotchy patterned cloud. They warn of precipitation within approximately 10 to 20 hours of sighting, if winds blow steadily from the Northeast to the south. Cumulonimbus cloud, which tend to follow severe weather are associated with severe wind squalls, however rarely predict a squall the size of a tornado. These clouds are also associated with hail, heavy precipitation and thunderstorms.
Stratocumulus clouds are generally found under a large storm cloud. They develop when waves form at the top layer of cold air, on which the clouds ride. They are an immediate indicator of bad weather from a sprinkle to a downpour. If at the head of a cold front then gusty winds or thunderstorms will follow. If located at sunset precipitation will follow within 12 to 20 hours, if the winds are from the Northeast to the south. Another type of cloud formation, Stratus Fratus, forms when small pockets of moist air manufactured by once fast moving thunderstorms.
We also noticed a variation within cloud formations. Vertically developing clouds seemed to carry with them a weather prediction all to themselves. Cumulus clouds that develop vertically soon develop into thunderstorm clouds. Cumulus Congestus cloud formations predict wind squalls and thunderstorms within five hours. Cumulonimbus, vertically developed, brings damaging winds, hail, and tornados. These clouds are also known as anvil clouds. We discovered that with close investigation and observation cloud formations can work as a trustworthy indicator of future weather.
We chose to use a statistical test for numerous reasons. For one, we were testing the statistics of the weather so it only seemed fitting to use a test that analyzed information on a statistical level. Also, we wanted to use a test that would give us more than just an accuracy percentage so we could look at the average levels for each phenomenon over the month that we took data. We also wanted to have a test which we could use with the information that we learned this year that dealt with t-testing and the null hypothesis, so Stat view’s unpaired comparison test was deemed the most worthy.
Included in the next section are our hard-copy results. They are composed of (1) the Stat View data table with information from October 10th through November 11th followed by the t-test of predicted readings to actual readings, (2) a comparative graph that puts each set of data into a visual context follows each data table. Finally, comparative graphs near the end of the packet show how we found relations between certain types of weather phenomenon.
Conclusion In conclusion to our project, we feel that we were quite successful in reaching our goals. Even due to the fact that we had a very limited arsenal of tools to use in our investigations, we still completed the statistical testing with a good feeling about what we had accomplished. When reflecting on our results, I feel that our results were the culmination of our research into the area of weather. We studied long and hard to hopefully understand why weather was the way it was, and our results show that we did, in fact, come to learn somewhat how the weather functions. After looking at our statistical testing results, one can see that, over time, we became very efficient at predicting the weather patterns and had a very good idea of where the readings for the future days would be. Our work fits in well with what others have done and what they are still doing. Weather prediction is a profession that can probably never be “perfected”, which means that there will always be room for improvement. Others, just as we have done in this project, will always be striving to better themselves on the subject of predicting the weather. This was a learning experience for each member of the group, just as it is surely a learning experience every day in the life of an actual meteorologist. I hope that we as a group continue to learn from what we have discovered about the weather in this lab and that we can use the information that we have come across in our everyday lives
Next Article
Previous Article
Return to the Topic Menu
Here is a list of responses that have been posted to this Study...
IMPORTANT: Press the Browser Reload button to view the latest contribution.
Respond to this Submission!
IMPORTANT: For each Response, make sure the title of the response is different than previous titles shown above!
Article complete. Click HERE
to return to the Natural Systems Menu.
Visit the rest of the site!