The Democrat and Chronicle's cover- age of rail-crossing safety began ear- lier this year after a retired couple were struck and killed by a CSX freight train as they drove onto a crossing in a Roch- ester suburb. Within a few days, it emerged that CSX personnel had purposely deactivated circuitry that controls the gates and warning lights at that crossing after re- peated complaints that the gates had become stuck in the down position.
The circuit was off-line for a full week without repairs made. During that time, crews were given orders to stop their trains before entering the crossing, dis- mount, make sure car traffic had stopped and then proceed. On the morning of the accident, the crew of a 110-car freight train failed to do SO. I was among several staff members asked to follow the story. As I did, I made frequent use of railroad safety data that is available online from the Federal Railroad Administration.
While the data has the usual accu- racy and timeliness limitations, it is as useful as any government cache I've seen for doing quick-and-dirty research. It's much like the Federal Aviation Ad- ministration airplane accident and maintenance data that many journal- readme.txt The IRE Resource Center offers copies of stories by journalists who have use computer-assisted reporting for transportation sto- ries.
Here are some examples: A Kansas City Star investigation found that the problem of fatigued driving among truck drivers is per- vasive. The Star analyzed truck- ing databases for the series. (Story No. 11812). An investigation by The (Cedar Rapids, Iowa) Gazette found semi-trailer trucks accounted for nearly half of all fatal accidents on Interstate 80 in Iowa. The Gazette analyzed the Fatality Analysis Reporting System (FARS) for the story.
(No. 21020). To order these stories and others, contact the Resource Center at 573-882-3364 or rescntr@ire.org and provide the story number. ists have used. The FRA online data can be tapped quickly to obtain back- ground in the aftermath of a train de- railment or highway crossing accident. It also can be employed to craft follow- up stories or to take a more thorough look at railroad safety. The data is available through the FRA Office of Safety Analysis Web site, http://safetydata.fra.dot.gov.
In contrast to the agency's main Web site (www.fra.dot.gov), which is largely be- reft of content useful to journalists, this site has nine separate sections rich with facts. I don't profess to be an expert on this data, but here's some of what you can find: information on highway crossing accidents, as well as the crossings themselves; train accidents (derail- ments, collisions of trains, etc.); inju- ries and deaths of employees and non- employees; and FRA inspections of equipment and crossings.
In our case, we-used the data to find other accidents on the rail line where the couple died, to look at the results of federal inspections of crossings in New York state and to pull out various nuggets of information about crossing accidents nationwide, Search and download Using the online query forms, you can pull off records of accidents or derailments in a few minutes' time. You can search by railroad, by state and county, by particular crossing and so on.
Some sections have 30 years' worth of data, though you'll find in some cases that you have to search a month at a time, which is highly annoying. There are literally dozens of different searches you can perform. The re- sults from these on-the-fly queries come mainly in the form of text docu- ments or pre-made text tables and charts. It's easy enough to paste such a table into Microsoft Excel or Ac- cess, but the text documents require a great deal of manipulation or data entry to get them into a database.
You can get some of this information by downloading entire databases, (see below). The summary reports of individual accidents that these queries produce are vague and at least occasionally inaccurate. The online summary of the Feb. 3 accident in suburban Rochester, for instance, continues to say that the "crossing was protected by gates, cantilever flashing lights, bells (audible)," but doesn't note that those devices failed to activate in a timely fashion.
You can obtain an Adobe Portable Document File of the crossing-acci- dent reports filed by the railroad by using the "Query and Generate Crossing Accident Reports" link. Hav- ing the ID number of the crossing in hand will help. You'll find that these more detailed reports also can be in- accurate. This summer, for example, I wrote about a 2001 fatal crossing accident that occurred in the Roch- ester area.
The railroad report I down- loaded indicated that the crossing sig- nals worked as intended. It turned out that wasn't true, as demonstrated rather vividly by a security-camera video of the accident scene around which we built our story. The same information that is included in the PDF accident reports is available in a full database that can be down- loaded from the FRA Safety data Web site. Other databases of train accidents, casualties and highway crossings also can be obtained.
The databases are available in six for- mats, including Access, Excel and comma-delimited text. The files come compressed or uncompressed. I used the Access 2000 format, and the download was flawless. I chose the uncompressed version; it took about a minute to get the 1.5 MB da- tabase onto my machine, using a high-speed Internet connection. That 1.5 MB file, though, represented a year of crossing accidents nation- wide.
That's the primary problem you can download only one year at a time. I spent about an hour download- ing and merging the tables to as- semble a 22-year collection of high- way-crossing accident data for New York state. That compilation has about 1,400 records in it; five years of crossing accidents nationwide has about 16,000. Each record contains 100 fields of data about a given ac- cident, though there is a fair amount of repetition.
I worked mostly with the crossing-ac- cident database, and found it helpful for several stories. You can sort not only for railroad or location or time frame, but also for the reported cir- cumstances of the accident: car hits train or train hits car; type of crossing equipment; gender of driver; reported speed of train; and so on. I used the crossing-accident data in several ways. I separated all the cross- ing accidents on the 20-mile-long rail spur where the Feb.
3 accident OC- curred, which was useful for a story about the claim that the spur had too many crossings. Finding problems For a story on CSX's maintenance record, I used data about FRA inspec- tions of freight railroads in New York state. I found CSX had been cited for crossing defects during inspections much more often than its competitors. The inspection data isn't in a downloadable database so I did some on-the-fly queries, and pasted the re- sults into a spreadsheet.
I also pulled data on crossing accidents in New York, and reported that CSX had a higher number of accidents than other freight railroads. For a longer story about CSX's cross- ing maintenance practices and how they contributed to the Feb. 3 double- fatal crash, I used the database to find how many accidents nationwide had occurred at crossings protected by gates and lights, and in how many of those accidents there had been alle- gations that the crossing equipment malfunctioned.
For a two-part story on the 2001 fa- tal accident, I was able to pull data showing how many accidents OC- curred when motor vehicles drove into the side of trains, and how many occurred at night. Both factors played a role in the accident I was chroni- cling, which occurred when the gates and lights didn't deploy promptly and the young motorist plowed into the side of a locomotive, possibly with- out ever seeing it.
There are literally dozens of different searches you can perform. Understanding data Before I worked with the data, I had to understand it by looking over the data- base file structure document, available for download from the Web site's down- load section. As I went through that file structure, I found myself a bit confused on sev- eral key points, and it wasn't until I also downloaded a second document that they made sense.
This document is a blank copy of the agency's "High- way-Rail Grade Crossing Accident/In- cident Report" form, which can be found at the very bottom of the main query page under FRA Forms. The blank form contains some of the codes used in the database, which are not replicated on the file struc- ture sheets. One example is the type of crossing equipment. Another and this is the one I was really after - is whether that equipment was working as intended when the acci- dent occurred.
So how do you find this in the data- base? I'm getting a bit arcane, but it took me days to puzzle out on my own, and it might be worth filing it away just in case you need it. In the main crossing-accident data- base are two fields entitled "Signal" and "Sigwarnx." If the crossing in question has automated equipment such as gates and lights, then the "Signal" field should include a digit from 1 to 7.
But what do those digits represent? If you consult the instruction page that comes with the accident/incident re- porting form, you'll discover that those digits represent the reporting railroad's statement about whether the gates and lights worked properly. "1" indicates everything worked well; anything else suggests an alleged or confirmed problem. Go one layer deeper and you'll find that if the railroad reported "5," "6" or "7," which stand for "confirmed" sig- nal problems, they also must fill out the "Sigwarnx" field.
This field, which uses letters instead of numbers, gives 19 possible explanations for the fail- ure of the crossing equipment. The explanations include things such as "devices down for repair," "vandalism," or a train going too fast for the cir- cuitry to keep up. (Here's a story tip for someone: Look in the database and see how often "insulated rail ve- hicle" is the explanation for a cross- ing-signal failure.
It turns out that some railroad maintenance vehicles are made in such a way that they won't activate gates and lights.) Admittedly, crossing activation prob- lems are rare - or, if you believe the rail critics I spoke with, railroads only report such problems rarely. But if you're writing about a crossing signal failure as I did in two local cases, it is immensely helpful to be able to pull out other accidents that might have occurred under similar cir- cumstances.
Contact Steve Orr by e-mail at sorr@democratandchronicle.com.