Tornado History Project - Over 53,000 tornado maps. US tornadoes from 1950-2009.

Just a quick note to document a few changes to the database where the SPC has modified data for past years:

• 19710812.32.1 - county changed from White Pine to Clark
• 20000213.13.3 - number of fatalities changed from 6 to 7
• 20061225.12.42 - Fujita rating changed from 0 to 2

A number of changes were made to 2008 records:

• 20081227.17.1688 - removed
• 20081227.48.1689 - removed
• 20080522.56.32767 - added
• 20080711.55.32767 - added
• 20080728.46.32767 - added

As a result of the modifications to 2008 data, the count of tornadoes for the year is now 1690 (from 1689), while the total count since 1950 is 53254 (from 53253.)

I am pleased to announce that 2009 tornado data has been added to the database. 1,156 tornadoes were added this year, bringing the total number of tornadoes in the database to 53,253. Just visit the front page to get started.

Note that, at the time of writing, the SPC has not assigned sequential national or state tornado numbers. However, they are using temporary non-sequential tornado numbers for the national number, which has been carried over here. Once the SPC officially sequences the tornadoes, our data will be updated.

In a previous post, I noted that there were some significant revisions to the 2006 and 2007 SPC historical tornado archive (the source of data for this site.)  According to Gregorgy Carbin, the Warning Coordination Meteorologist at the SPC, the National Weather Service changed the way they reported latitude and longitude values in October of 2006.  The person responsible for maintaining the SPC archive at the time inadvertently continued to convert the lat/lon values from minutes/seconds to decimal degrees, even though they were now being reported by the NWS in decimal degrees.  This unnecessary conversion caused the lat/lon values to be slightly off.  When this was discovered in late 2008, he corrected the lat/lon values, while also taking the opportunity to do some additional quality control.

As of last Saturday night (10/10/2009), these changes have been incorporated into the Tornado History Project database.  Although not a comprehensive list, some key changes are listed below:

• The number of tornadoes for 2006 now stands at 1103 (up from 1101)
• The number of tornadoes for 2007 now stands at 1098 (up from 1093)
• 4 tornadoes were added for 2006 (20061230.22.42, 20061116.37.31, 20061229.48.115, 20061229.48.116), while 2 were deleted (20060311.5.7 and 20060311.29.17)*
• 11 tornadoes were added for 2007 (20070826.38.39, 20070914.37.1, 20070914.37.13, 20070920.27.19, 20070928.20.114, 20070928.20.115, 20070928.20.116, 20070928.20.117, 20070928.20.118, 20071019.26.24, 20071227.48.20), while 6 tornadoes were deleted (20070328.31.30, 20070504.20.41, 20070505.46.50, 20070505.46.59, 20070914.37.9, 20071018.26.19)*

* Index numbers are generated by the Tornado History Project to uniquely identify tornadoes within the site. The index number is not part of the official SPC data.

I am pleased to announce the addition of the 2008 tornado data from the Storm Prediction Center (SPC). Simply go to the search page to have a look.

I should note that the SPC has not assigned state tornado numbers this year (as of the time of writing.) Although a minor issue, this does affect how tornado index numbers* were chosen for tornadoes. In past years, the index number was generally a mix of the date, state FIPS code, and state tornado number (see “2006 Data Added to Tornado History Project Database”. However, since the state tornado number has not yet been assigned, the general SPC tornado number was used instead.

One more note on the archive files posted by the SPC this year…I noticed some significant revisions to the 2007 and 2006 data (mainly tornado number and latitude/longitude revisions), but those changes have not yet been incorporated here. So, be on the lookout for another update to address this in the near future, as well as several other general updates to the Tornado History Project.

* Index numbers are generated by the Tornado History Project to uniquely identify tornadoes within the site. The index number is not part of the official SPC data.

One of my favorite recently added features of the site (see Major Upgrade to Tornado History Project 7/19/08) is the ability to easily correlate each tornado in the Tornado History Project database to the original line (or lines) in the Storm Prediction Center (SPC) archive. Why is this important? It is not uncommon for users to email me with a comment such as:

The county listed for tornado X is wrong. It should be Y - I know because it hit my house. Why don’t you do a little research before putting this on the internet?.

Yikes! While I want to know when there is a suspected error in the data, I also want to be clear where the information comes from. So, this will make it much easier to tell if the Tornado History Project is accurately reporting the data as given in the original SPC archive. In short, it adds transparency to the site and takes the mystery out of where the data comes from. And in cases where the Tornado History Project has changed any of the data (which is very rare), notice is given.

Let’s take a look at where you can find this info. On any individual tornado map page, all stats for that tornado are shown in an easy to read format immediately under the map as shown in the picture below. The last line of the section shows the actual text from the SPC archive used to generate those stats.

Of course, being able to actually see the original line from the SPC archive isn’t quite enough. It is also important to know how to read and interpret the information. So, a link is also given to the SPC instructions that explains what each piece of information actually means.

I should point out that there is one situation in which the approximate path shown for a specific tornado may not actually be based on the data in the SPC archive. Please see a previous blog entry (Enhanced Paths Explained 7/19/08) for more information.

And finally, there is one more note on this…Users must be logged in to see the corresponding line (or lines) from the SPC archive, so if you don’t see anything it, just login or register for a free account and you’ll be all set.

One of the most exciting new features of the site (see Major Upgrade to Tornado History Project - 7/19/2008) is the inclusion of enhanced path data derived from the National Climatic Data Center (NCDC) historical archive (thanks to Stuart Hinson for providing access to the archive!) While there are many differences between the Storm Prediction Center (SPC) historical archive (the primary source for the site) and the NCDC archive, the most significant difference is the method used to give coordinates.

In a nutshell, for tornadoes in which path data is available, the NCDC archive gives latitude and longitude coordinates when a tornado enters another county, while the SPC archive only gives coordinates when a tornado enters another state. This means that multi-county tornadoes in the NCDC database may have more data points available to draw a more detailed approximate path on the map. Thus, instead of the straight lines generated for SPC derived paths, NCDC derived paths generally have more turns and zig-zags.

How big of a difference does the inclusion of NCDC data make? Well, that depends on the number of NCDC derived data points given for any specific tornado, but let’s look at an example. The screenshot below shows a few F5 tornado paths using SPC data. As you can see, the approximate paths are straight lines, except for where a tornado crosses a state line.

The screenshot below shows the same area using NCDC derived data, when available. As you can see, some of the paths show a greater amount of detail.  Compare the apparent path near Jackson, Mississippi, for example.  With the enhanced path data below, the path passes the city from the South rather than the North.  And the more you zoom in, the more the difference will become apparent.

Displaying enhanced path data was not an easy task and took me almost a year to complete. The main issue was that there was no easy way to match up the tornado records between the two historical archives. I had to do quite a bit of manual tweaking to get it to work. Nevertheless, I think it was well worth the effort. I haven’t done an exact count, but I estimate that there are roughly 2,000-3000 tornadoes with enhanced path data, out of the 50,000 tornadoes in the database (the vast majority of tornado paths are short and only affect a single county.) Please note that you can only see enhanced path data when you are logged in.

I believe this to be the first time anything like this has been done. If someone knows different, please let me know. In any case, I hope you will sign up and do your own tornado search.

It has been a long time coming and is long overdue, but I have finally released a significant update to the Tornado History Project. For those who have been here before, the first thing you might notice is a change to the general look of the site. In short, it just looks better. But that is minor compared to the new features. I’ll post more in-depth details on some of the new features later, but the major ones include:

• Historical data extended through 2007
• A user account system
• Enhanced tornado path data for select tornadoes (see Enhanced Paths Explained 7/19/2008)
• Ability to correlate any tornado with the related line in the original SPC archive
• Ability to export custom data searches
• New buttons on the map, including experimental integration with Google Earth
• Improved functionality in the tornado forums (there are over 50,000 forums available.) Improvements include:
  • Immediate posting of user comments (no need to wait for them to be approved)
  • Ability to watch any tornado forum for comments
  • A user ranking system (F0 through F5)
• Other minor improvements

While these improvements are important, their real value is that they are the necessary foundation for future development. Keep checking back as I hope to improve the site even further. And remember, the purpose of this site is to recreate the history of as many tornadoes as possible through eyewitness accounts, personal memories, photos, and videos. I can’t do that alone - I need your help. So, tell your friends about this site, post your own memories or photos and videos, and, if you can, make a small donation.

It recently came to my attention that the time given for tornadoes occurring in 2004 was incorrect. The root of the problem is that the time as given in the source data (SPC historical archive) are in a different format for 2004 than in all other years. Thus, the standard translation used for all other years does not work for 2004. To make a long story short, the Tornado History Project database has been updated with corrected 2004 tornado times.

The latest update to the Tornado History Project has made searching for tornadoes easier than ever. Previously, there were separate pages for searching the database and outputting the results on a map or in a table. So, if you were viewing the search results in table form, but wanted to view the same search on a map, you had to 1) go to the map page - which erased your previous search parameters 2) re-enter your search parameters 3) submit your search. Now, a user can simply use the combined search page and toggle between the map and table outputs.

Also, in previous posts I have mentioned that improved path data was forthcoming for selected tornadoes. This is still in the works, but has not yet been completed. I will post here when it is ready to go.

The 2006 data has finally been added! I know I initally said I’d be done with the 2006 update by the end of September, but it proved to be more difficult than I had anticipated. Well, actually, getting the data added wasn’t the difficult part - it was the near complete internal database reorganization I decided to do. To make a long story short, there are several inconsistencies within the SPC archive. I’m not talking about errors in the data such as incorrect Fujita classificaions, lat/lon coordinates, etc. (there are still many of those…), but rather, inconsistencies with how segments of the data relate to one another. I have identified most, if not all, of the inconsistencies and have made modifications where needed. The modifications were generally limited to the internal SPC numbering scheme (the “SPC Tor. #” and “State Tor. #” columns on the Tornado Database page) and do not affect the official stats such as locations, fatalites, injuries, etc. Any modified records will be displayed in a red font when viewing search results on the Tornado Database page.

Also, significantly, I have completely changed the internal tornado indexing scheme. This shouldn’t affect anything, but it is an important change in that the new index will serve to permanently identify tornadoes in the Tornado History Project database going forward. The index is derived by combining the date of the tornado, the FIPS code of the state it occurred in (or the first state it occurred in if it is a multi-state tornado), and the state tornado number of that tornado. So, a tornado that occurred on 1/1/1950 in Alabama (FIPS code=1) and has been given the state tornado number label of 1 by the SPC will become tornado index #19500101.1.1
In addition to internal database changes, I have made the site somewhat easier to use (I think…) by providing clearer instructions and more appropriate icons representing links to maps, user comments, photos, and videos. You can see this by taking a look at the Tornado Database page. Tornado paths have also been changed to match the color of the tornado icon. So, an F3 tornado is now represented by the old orange icon and the new orange path line (instead of a standard blue path line…)

A few other changes were made too, but probably not significant enough to discuss here. Keep checking back over the next few weeks/months as I have a few other updates planned. The biggest update will be adding more detailed path data for many tornadoes (see previous blog post for more information.