To encourage people to evacuate, 3DNature’s Chris Hanson created an animation predicting what could happen if the storm hit Galveston, Texas.

3D Nature created the terrain visualization software, Visual Nature Studio, used to model, render and animate natural and manmade environments with total photorealism. The company’s programmers believed they were uniquely qualified to use publicly available data to create a visualization of New Orleans after Hurricane Katrina that would express visually, geographically and intuitively what actually happened and when. Besides creating the visualization as a public service, made available to new agencies for downloading from the web, the animation demonstrated the capabilities of their software. Chris Hanson, one of the company’s lead programmers says, “Nobody called us up and asked us or paid us to do it. We felt that it needed to be done to fill an information and understanding vacuum about the whole event.”

After the storm, programmer Frank Weed created 3D visualizations of the progression of the water level in New Orleans caused by Katrina — an accurate detailed representation of how the water inundated the area. They also attempted to determine why one area flooded when another didn’t. Timeline, water height and dike height info was gathered from various sources, including the Army Corps of Engineers. Although they were not an official government agency, they were able to acquire good terrain model data that showed every little bump, even large buildings in the city.

3DNature was able to secure high-quality data resources to visualize New Orleans accurately during Hurricane Rita. The Galveston animation lacks local details.

With Hurricane Rita bearing down on the heels of Katrina, Hanson realized the potential for an enormous amount of damage to Galveston, Texas. With the view to encouraging people to evacuate before the storm hit, two days before the projected landfall Hanson created an animation predicting what could happen if a storm surge of up to 20 feet topped the 17-foot retaining walls. Fortunately, the storm veered east and hit Galveston on the best possible front, an area of marshlands heavily protected by a seawall — features represented in the 3D visualization. Hanson, who had been following the television coverage, concludes the city, “dodged the bullet.”

With the Galveston animation, they rushed to complete the visualization before the storm hit. Without the clout of a government-sponsored organization behind them, they were unable to get data resources of as high a quality as those they secured for New Orleans. The island depiction of Galveston is about the right shape and in the right place, but it doesn’t show precise local details. Hanson explains, “We have done this for 13 years. During that time we made a lot of bookmarks about where to find data. In the geospatial world data is the first and worst problem. It can be very expensive to obtain if it isn’t already produced. If already produced the quality can be good or awful. It’s worse outside the United States because the U.S. government through various departments produces terrain and other geospatial data for their own programs. They make it available to the public at nominal cost.”

An example of SDSU’s imagery data as “underlayers” on which large amounts of other data, like hospitals, people, clinics, refugee sites, are placed.

Most of the New Orleans terrain model data used was publicly available and downloaded from the USGS website. It was created using LiDAR — light detection and ranging technology. The terrain photos were made from a scanning laser mounted in the belly of a plane. The laser sweeps a swath underneath the plane. Thousands of times a second the laser scans the landscape while firing a short coded pulse of light that counts until that pulse is reflected back to the plane a few nanoseconds later. The time difference between when it left and when it returned makes it possible to calculate how far below the plane that spot is located. The plane carries GPS, making it is possible to ascertain where the plane is geographically at any given time so they can calculate where that laser hit. They cover an area recording these laser hits. With enough of them a surface can be mapped. Typically, at the same time that the laser scan is made, digital photos or film is also shot, then integrated with the scanned data to give not only the terrain shape, but also the appearance of the surface at the same spot.

3D Nature’s software accepts a huge number of file formats and converts them into a usable format. The terrain data came in latitude and longitude, which allows the program to position the data in the real world. The image data had a different set of coordinates, but the software enabled it to line up with the terrain. The team gathered timetable and other event information such as water level from whatever sources they could find — news accounts, water monitoring stations and NOAA. They extrapolated a centralized list and timetable of events. When that was complete they had a solid idea of what had happened and when. Their task was then to take the dry model that they had created and add water. The software lake component allowed them to outline an area and control the placement and height of water. Textures and reflections were added for greater realism. They animated the speed of the flooding to a timeline determined by various written interviews and media accounts. The assumptions made in the animations were proven by additional data captured by NOAA post-storm.

Mary Ann Skweres is a filmmaker and freelance writer. She has worked extensively in feature film and documentary post-production with credits as a picture editor and visual effects assistant. She is a member of the Motion Picture Editors Guild.