Around this time last year, I stumbled on IbisEye, an online hurricane application on the Web site of the Sarasota (Fla.) Herald-Tribune. I described it as a perfect example of what Adrian Holovaty calls “journalism through computer programming.” In its first year online, IbisEye won a Knight-Batten Award of Distinction.
“We were blown away by this amazing tool and database that also solicits the wisdom of the crowd,” the judges said.
With the onset of this year’s hurricane season, I asked the person who came up with the idea of IbisEye, investigative reporter Maurice Tamman, to discuss its roots, how it works and what it provides site visitors.
Why did the Herald-Tribune create IbisEye?
Living in Florida is living with hurricanes. After Charley hit the southern end of our coverage area there was an evolutionary process of bringing together many data sources to create a unique application to deal with tropical storm season.
What’s the origin of its name?
Legend has it (at least University of Miami Hurricanes legend) that the bird is the last to leave before a hurricane hits and the first to return. It reminded us of what reporters do and we hope the site informs and illuminates online users in a similar way to how print reportage traditionally does.
This is IbisEye version 2.0. What have you changed and what lessons lie behind those changes?
The entire site was built up from scratch. We consider last year to be beta and this year v1. Our primary goal was to make the application as fast as possible. Last year’s version was cool but damn slow. Many of the same features return but they are all improved and much easier to find.
I think we’ve succeeded.
In addition, we found a much better database of windfields and we’ve significantly improved how we display those windfields. We also added calculations for the southeastern population and cities exposed to hurricane force winds along with our analysis of Florida property exposed to hurricane force winds.
The alert system is also much improved.
The damage reporting system works for any area hit by hurricanes: search for an address, click on the icon and report damage. All the reports are tied to a latitude and longitude or a parcel ID number (in Florida).
We’ve also added an alert system that can either be tied to a parcel or a point on the map (or a weather alert zone); anytime a selected alert is issued an e-mail is sent out. In addition, the system (and we haven’t tested this yet) should send out alerts if the point is exposed to hurricane force winds, is forecast to be exposed or is inside the “cone of death.”
We also expect to add alerts for people who want reports of damage or lost people and pets near a particular point.
There is more, but that’ll do.
What is the value of IbisEye to consumers?
It’s meant to be a wandering experience; come in look around; learn. And, perhaps more importantly, it’s designed to give context and tools for dealing with and evaluating hurricanes.
How wide has its reach been?
The system is designed to be almost fully functionally throughout the Southeast and especially in Florida.
How much time did it take to produce versions one and two?
In both cases, we spent about six weeks building the application. Last year, however, we added and fiddled, and subtracted and multiplied all season long.
We have two or three more features to add and then we’ll be done for the year.
What skill sets were needed to create it? Who was on the team and what did each contribute?
There are two of us and it really was a partnership.
Charlie Szymanski handled all the client-side scripting and design implementation. I handled the server-side programming and the data management. The new look is largely his doing.
Charlie has a fine eye for Javascript and ActionScript (flex/flash) and without him, all this data I put together would be unusable. But you do need someone with decent knowledge of server-side programming and database management.
How did you get the information required to build IbisEye?
The historic storm data came from a professor at Colorado State. Florida parcel data came from the Department of Revenue and county GIS/property appraiser departments. There is a smattering of census data and a whole bunch of geo-spatial data from scores of sources.
We also use live weather alert feeds from Weather Underground, who also supplies us with the current season storm data. We get weather station/buoy data from WeatherFlow.com.
We’re also adding some satellite images, which we grab directly from NOAA.
What, if any, barriers did you face obtaining that info?
The property parcel data took a little leg work, especially the parcel level maps. Other than that, the really heavy lifting was done once we got the data and set up the various scripts to generate all the reports.
It should also be noted that the Javascript work Charlie did was massive and extremely complex — he might choose other words.
What programs did you use to create IbisEye?
IbisEye is a .NET application which we developed using MS Visual Studio. The data sits on a couple of SQL Servers. We also used ArcGIS and Photoshop and a few other custom applications.
IbisEye is ideal for Florida’s hurricane season. Can you envision similar tools that news organizations facing different natural disasters could create?
Our damage reporting system/alert system, could be adapted in a few minutes for virtually any natural disaster in the U.S. and Europe. (The address lookup for other parts of the world aren’t so hot.)
And the weather alerts and winds are independent of active storms.
A bottom-line question: What’s your evidence that IbisEye was worth all the trouble?
Nothing really tangible — yet. Last year was all about proving we could build the system.
It’s as much about trying to think about news in creative and dynamic ways rather than rushing around (like so many newsrooms) just trying to follow the next fad, whether it’s (usually boring) online video that no one will watch, or (too-often-uninformed) blogs or worse, uninformed citizen journos.
Our job has always been to bring clarity and context while also serving as a useful, educational and entertaining tool. That, however, is for others to judge.
I do know that for those brief moments when we had storms last year, hits to the site exploded. I expect the same this year.
We’ve also developed a system for overlaying hurricane-related services/businesses tied to the bounds of the map with an eye toward developing some kind of business model.
We categorized services based on their practical use to the public given the tropical storm status for an area. For example, fire- and flood-damage cleaning services would only show up after a storm passed through an area while hurricane shutters installers would only show while no storms threaten.
Right now, the services are only available in our coverage area but we could expand to all of Florida, or even the Southeast if there was the money to support.
And we’re ready to make those services searchable if we expand it.
Was it worth it? I don’t know. Ask me in five years. If I’m still in the biz, then I’d probably say yes.
What was the biggest surprise of creating IbisEye v1 and v2?
How bad last year’s version was. I can barely look at it without cringing. The code makes me puke.
What were the most important lessons you and your staff learned from the project?
I assume this is a big-picture question. All I can say is what I’ve know for years. Focus and dedication to a task produce results. Talking and meetings result in nothing. But I don’t need to say that, we’re all the ink biz.
What would a newsroom need to create a project of this magnitude, in terms of hardware, software and wetware?
Time. Charlie and I toiled untold hours every week, nearly every day for six weeks. Sometimes a small group of people dedicated to one task can finish something faster than a division of programmers and managers.
We run the application on three servers; all windows boxes. We run them outside of the corporate structure, which was essential. The rest I think we’ve covered.
The cost is relatively small, in the grand scheme of things, and some of that cost (such as the cost of the servers) has been used for many other projects.
Are there any questions, subjects, issues that I haven’t addressed that you think readers should know about?
Strangely, I’m torn. I certainly don’t want a busy hurricane season but this application only really takes off when it is busy. There is a part of me that wants to watch the wheels turn.
