Gone are the days when the phrase “educational TV” would inevitably send shudders of dread through kids and teens. Quite the contrary, many of today’s educational programs are fast-paced, expertly narrated and full of surprising, fun, and visually engaging facts and science trivia. But while no doubt entertaining, are these programs still any good at their core function, namely teaching the kids actual concepts useful for understanding modern science?
The long-running PBS educational series, NOVA, aims to satisfy this goal by providing not only standard teachers’ notes and lesson plans to accompany its shows, but also by investing in extensive and realistic online laboratories in which students can explore actual scientific datasets. A good example of this approach is provided by the NOVA Cloud Lab, which is actually far more exciting than it sounds, as one of its main sections concerns the formation and study of storms. There are many neat parts to NOVA’s Cloud Lab. You’ll find sections that explain various aspects of cloud formation, structure, and effect on weather and climate. The centerpiece of the cloud lab, though, is the research activity appropriately named Reconstructing a Storm. Weather is incredibly complicated. Our ways of gathering data on weather are sophisticated, but by the very nature of what is being studied, put out a lot of data that is not always easy to interpret. In the last half century, humankind has put various remote sensing weather observatories in orbit, and a significant part of the everyday work of meteorology and climate science is knowing how to interpret the streams of data we get every minute from these orbital platforms. Dealing with vast quantities of diverse environmental data, called “Big Data”, is one of the principal challenges of modern science.
Since it’s a relatively new development, however, dealing with Big Data has not yet become a priority in education. Interactive features like NOVA’s Reconstructing a Storm aim to bridge this vital gap in science education by providing students a taste of the challenges and rewards associated with analyzing Big Data.
Part one, Inside a Megastorm, uses the example of Hurricane/Super Storm Sandy to introduce students to the various kinds of data visualizations as they relate to tracking a giant storm. Since the overall dataset related to Sandy is huge and messy, this part of the lab focuses on several distinct phases of the storm, introducing relevant concepts and data visualiations to each. Thus, as the storm began to develop over the Caribbean, one of the featured modules takes on unusually cold cloud top temperatures.
The cloud top temperatures dataset is a perfect example of the kind of thing that atmospheric scientists deal with daily, but that is usually omitted from any public discussion. After all, why would anybody care what the temperature is at the top of a cloud? It turns out that when weather satellites look down on our atmosphere, they “see” in various “colors” of the infrared part of the spectrum, and can reconstruct the temperature at various heights in the atmosphere from these observations. However, clouds are opaque to infrared just as they are to visible light. So the lowest that a satellite can “see” in a cloudy scene is precisely the cloud top. And the temperature of these tops of clouds tell us, roughly, how high up these clouds are, since air temperatures in the stratosphere are closely tied to height above sea level. A storm whose highest cloud reaches 8km (atmospheric scientists like the metric system) isn’t likely to be anywhere near as severe as one with towering 16km clouds.
Students don’t just learn facts like this through some artificially designed classroom scenario. They see the same data that scientists saw when they were predicting the path Sandy would eventually take.
Part two, Analysis and Reconstruction, gives the students several new storms to examine, and hands over some control in the form of guided data interpretation exercises. Big Data isn’t just voluminous, it’s also messy – full of noise, artifacts derived from imperfect observation techniques, seemingly irrelevant detail. A map of something like cloud top temperatures over a given geographic region isn’t anywhere near as easy to read as the cleaned-up google map satellite overlays we’re used to. For one thing, the data is only available on those pixels in a scene where a cloud is present at the time of observation, so large parts of the scene have no useful data at all. There’s a color scale presented as a legend next to the visualization, and it’s the student’s task to answer some simple questions meant to probe their understanding of what’s being shown. How cold do the cloud tops get in this scene? What is the maximum or average estimated precipitation in that one? Students get used to looking at satellite data not just as a pretty picture, but as data that can be read and interpreted.
Next, students are asked to place the observations they just analyzed in context. Specifically, their task is to take some of the above-described datasets and predict which part of a particular storm’s track they fit into. This exercise is both challenging and critical to the core educational aim of Reconstructing a Storm. It shows by example how individual datasets are only a small part of a bigger picture, that can only start to emerge once each piece of data has been properly understood in isolation.
Once students have had a go at guided data interpretation, they’re unleashed on what is the probably the riskiest, as well as most ambitious part of the project. Open Investigation is exactly what it sounds like: Many years’ worth of observational data from weather satellites, viewable a day at a time, presented so that students can overlay layers of experimental data. While the focus here is on unstructured learning – even today’s data can be seen as soon as the satellite data processing centers release them, the data for dates and locations of notable storms are accessible from a pulldown menu. The student who selects one of these pre-screened weather events is presented with an analysis task specific to that storm: How did a low-pressure area change the storm’s track? How did landfall affect the storm’s precipitation? How does the temperature in the eye of the storm compare to that of the surrounding cloud wall?
Perhaps most impressively, the Cloud Lab is only one of several such interactive modules being designed by NOVA. Besides the existing cloud, sun and energy labs, NOVA is in the process of developing labs dealing with RNA, cybersecurity and the brain. Each of these areas depends heavily on Big Data for its everyday research. And NOVA labs is bridging the gap between traditional science education and the modern data-centered approach to science by helping students become comfortable with these data and their associated visualizations.