Time, Space, and Energy (The Future of Tornado Archive)

Hi! It’s Ava from the content team, and I’ve been so excited watching this project take shape. Despite being an all-volunteer effort, we’ve been able to do so much already. Every person assisting with the project has been integral to building what we have so far, and we’re not done yet. As of now, our tornado explorer is (thanks to our wonderful programmers) one of the quickest ways to explore the history of tornadoes, even when limited to the standard NWS dataset. However, as those who noticed pre-1950 tornadoes might have gathered, we’re not restricting ourselves so tightly.

Time

One of the greatest limitations of prior attempts at tornado archiving projects has been their timespan. The NWS’ tornado database spans back to 1950. Compared to the USGS dataset, which catalogs quakes back to 1900, and the HURDAT database, which has every known hurricane since 1851, this is a tiny temporal reach. So many notable tornado events are not captured by this data. The Tri-state tornado, the deadliest tornado in US history; the 1932 Deep South Tornado Outbreak, referred to as a super outbreak by NWS Birmingham; and the tornadoes that struck Tinker AFB, heralding the first tornado forecast, are all absent. This deficiency was the first thing we wanted to address.

It’s not like this data doesn’t exist. Thomas P. Grazulis, to whom we owe a huge amount of thanks for compiling the data (and making it public domain, no less), constructed a massive book of every significant tornado in the US since 1880. Unfortunately, this book’s tornadoes didn’t come with starting and ending coordinates. Instead, each description contained a general path relative to surrounding towns. Think “from 5 miles west of Norman to 8 miles east of Oklahoma City.” Figuring where to place the endpoints from this description would be simple. However, others didn’t give distances, some didn’t give directions, and some listed towns that no longer exist. Determining tornado paths from this is as much an art as a science. So, a caveat: DO NOT treat the pre-1950 data as if it is particularly accurate. The content team has done its best to get these tornado paths as close to their descriptions as possible, but there will be errors, possibly large ones. Be careful. 

This task of converting from relative to absolute locations is also why we only have violent tornadoes from 1880-1949 as of now. Over half of the F3s and a good percentage of the F2s are completed, but there’s still a lot to get through. Implementing the rest of the significant tornadoes after 1880 is an ongoing process, but it should happen. Eventually. 

In the further future, we intend to integrate the DAT tornado data from the NWS too. This data includes damage polygons and much more accurate tornado tracks. If you want a sneak peek at what this data will look like, the El Reno tornado on May 31, 2013, is an example of the level of precision to expect. However, aligning the DAT data with the current database will probably be labor-intensive, so high-res tracks won’t be more common on the map anytime soon.

Space

Once we started expanding our dataset in one area, we inevitably started thinking of other shortcomings. While the US data can be frustrating to use, other countries have it worse. There are very few places with any tornado exploration tools at all, and there’s been no effort (as far as I can tell) to unite international tornado databases into one place. Filling this gap is our next goal. We’ve already found databases (at various levels of usability) for Canada, Japan, South Africa, Australia, and Bangladesh, and at some point, we hope to implement all of these. Additionally, we’re working to access even more countries’ data, but that plan might fall through, so no promises (12/23 UPDATE: The European Severe Weather Database has given us permission to include their database of European tornadoes). The task of implementing these tornadoes is not one without difficulty. Many are in different formats, are missing data, or have weird mistakes. However, expect more than just the US to be searchable.

Energy 

Convective Available Potential Energy, that is. There are several atmospheric reanalyses that fill in the gaps left by in-situ observations, and tools with which one can access them. However, they can often be difficult to navigate, and comparing the dataset of historical tornadoes with reanalyses can be frustrating. Having easy access to this data would be endlessly helpful, as understanding the environment that led to past tornado events is both historically important, and useful in learning how to forecast future severe threats. Thus, v2 will include a couple of features to make this search easier.

One of the easiest ways to get a quick handle on atmospheric conditions is to view a Skew-T diagram, also called an atmospheric sounding. We’re unsure how soundings will be generated yet, either through point-and-click generation as found in existing tools (https://atlas.niu.edu/narr/, for example), or by generating soundings automatically for each tornado in the database. Either way, these plots will be a fantastic resource to have when looking at past tornado events.

Last but not least is the feature that I’m personally the most excited about. Tools for viewing maps of reanalysis data are pretty sparse and form the biggest absence in weather archiving websites. Which is where the environment browser will come in. At the bottom of the explorer page, a fourth tab will be added, where one can select and view environmental parameters as a map underlay. Once the desired parameters are selected, a user can then press the play button, and watch an hourly animation of both tornadoes and atmospheric conditions within that time frame. This feature is not likely to be available for all days (the target is all violent tornado days + other notable days such as May 31, 2013), but when it’s accessible, it’s awesome. An example of what the map looks like with environment data is below.

With all of these features, we hope to be greatly beneficial to the study of tornadoes as a whole. This launch is a step toward an easy-to-use yet feature-rich way to learn more about these devastating storms. I can’t wait to share more in the future as these additions become reality.

Summary of planned features (Not necessarily in v2):

• All significant tornadoes 1880-1949
• High-res and polygonal tornado tracks from DAT
• Non-US tornadoes (Canada, Japan, South Africa, Australia, and Bangladesh, or more)
• Skew-T viewer
• Environment viewer for violent/notable tornado days