A Coast Guard helicopter flies over the flooded area in New Orleans on Sept. 4, 2005, looking for survivors of Hurricane Katrina. Photo by Liz Roll, courtesy of FEMA.

Editors' Note: This article was submitted to POB months prior to Hurricane Katrina's landfall and the subsequent destruction in the Gulf Coast region. The authors revised this article shortly before it went to press to address this project's application to Hurricane Katrina. We have published this story because it demonstrates the importance of the National Geodetic Survey's subsidence studies, and the significance of elevation models for evacuation routes in coastal cities and for the rebuilding efforts to come. Our hearts go out to all those devastated by the hurricane.

In early August 2005, the National Oceanic and Atmospheric Administration (NOAA) issued a hurricane season outlook that included in total, a season "likely to yield 18 to 21 tropical storms, with nine to 11 becoming hurricanes, including five to seven major hurricanes." Hurricane Katrina proved to be one of these dreaded storms. As the southern part of the state was only a few feet above sea level and sinking in many places, it was more than susceptible to devastating floods.

Residents facing Katrina were mandated to evacuate because of the incredible magnitude of the storm. But there will be other less powerful storms in the future that may or may not cause massive flooding. The decision of whether to evacuate, as well as when to evacuate, will need to be made by well-informed disaster managers. To aid these managers in their future decisions, NOAA's National Geodetic Survey (NGS) has been conducting a project to measure highly accurate road elevations of the evacuation routes in Louisiana.

Subsidence in Louisiana

The need for accurate elevation measurements in Louisiana has been complicated by subsidence, the slow sinking of the upper crust of the Earth. Historically, this sinking in Louisiana was offset by the accumulation of silt carried into the area by the Mississippi River. In the early 20th century, Louisiana actually grew in size as new land was created by sediment deposits from the river.

By the end of the 20th century however, that process had been halted. The Mississippi River was channeled, levees were built and the river was contained inside them. The river no longer flooded its banks, and its silt is now sent over the continental shelf into the deep ocean. Although nobody can say exactly why subsidence is occurring in Louisiana, there are a number of hypotheses that include loading of the lithosphere (the earth's crust), soil compaction, oil extraction, ground water extraction, slow moving earthquakes and salt domes.

"We have known that Louisiana [has been] sinking for some time. I can remember reports [about it] back in the 1970s," says Dave Zilkoski, deputy director of the NGS. The reports that started to get attention were tidal records showing rapid sea level rise in Louisiana. In contrast to other tide gauges in the United States and around the world, the Louisiana tide gauges showed a rapid rise in sea level reportedly because the land was sinking. But it was not until the advent of GPS technologies, and the advances in accuracy in the vertical realm, that NGS was really able to begin to measure the rates at which subsidence was occurring. The NGS compiled decades of data in its research to document some of the rates of subsidence observed in southern Louisiana, and released this information in NOAA Technical Report 50. This report showed that in extreme cases, bench marks were losing up to 2 to 3 centimeters of elevation a year.

NGS employee Kendall Fancher runs a final check of the equipment during a survey in Plaquemines Parish, La.

Planning and Measuring Evacuation Routes

When hurricanes approach in the future, disaster managers will need to make decisions about whether to issue evacuation orders and when to issue them. A number of factors are examined in this decision-making process. One critical factor that must be examined is the point at which the hurricane evacuation routes will flood. To estimate this, managers need to know information about the storm, but they also need to know a basic piece of information: what is the elevation of the road? The answer to that question is complicated. A map may give a stated elevation, but how long ago was that map made? What datum was used? How much has the road subsided since then?

In the summer of 2004, the NGS set out to determine an efficient method for determining accurate road elevations. It was decided that the time and costs involved in using conventional survey techniques to measure the sinking roads could prove to be major obstacles to completing any large-scale project in southern Louisiana. Furthermore, given the nature of subsidence, these surveys might need to be repeated as soon as they were finished. To address these concerns, the NGS needed to find a system that could be used to efficiently update profile data for transportation routes. The NGS decided to evaluate the Position Orientation System for Land Vehicles (POS LV) manufactured by Applanix, a Trimble company located in Richmond Hill, Ontario.

The Applanix POS LV system incorporates data from an inertial measurement unit (IMU), two GPS receivers and a Distance Measurement Indicator (DMI). The system blends the data from all four sources and solves for a best-fit solution based on the strengths of the individual sensors. The primary source of positioning is the in-vehicle mounted IMU, which generates a representation of vehicle motion in all three axes to produce continuous position and orientation information. The GPS receivers (also mounted inside the vehicle) and DMI (mounted on the hub of the rear driver's side of the vehicle) provide correction information to the IMU to minimize the "drift" of the unit. GPS antennae are mounted atop the vehicle. The POS LV's software package, POSPac, allows for post processing of the GPS data and blends this reduced GPS data with data from the other sensors to produce a smoothed best-fit solution.

NGS selected the POS LV for evaluation because it demonstrated the potential to determine profile data accurate to the subdecimeter level, reduce the impact on traffic flow during data collection, maximize the amount of data that could be collected in a short time span, and acquire data in areas not suitable for stand-alone GPS surveying. After installing the POS LV system on a Chevrolet Surburban, the NGS task force determined the spatial relationship of all of the components relative to a vehicle reference point using a Leica Geosystems (Atlanta, Ga.) TC2002 total station. NGS employees established the vehicle reference point on the trailer receiver so that the relationship between the height of the road surface to the vehicle reference point could be easily measured. By determining this relationship between the system's components and by designating this point as the vehicle reference point in the POS software, the system solved for coordinates at this user-defined location.

The Test Area

NGS collected profile data for almost 1,000 miles of hurricane evacuation routes and about 12 miles of hurricane protection levees during this project. These evacuation routes were primarily located in southeastern and southwestern Louisiana, south of Interstate Highway 10 in low-lying areas. The NGS task force dedicated to measuring Louisiana's evacuation routes spent six weeks collecting and evaluating data obtained with the POS LV. Data was collected, reduced and validated to produce the final product intended for use by the Louisiana Department of Transportation, the U.S. Army Corps of Engineers, the local parish government emergency planning personnel and parish levee district managers.

NGS employee Steve Breidenbach establishes a validation point on top of a flood control levee near Golden Meadow, La.

Data Collection

In order to achieve a high level of accuracy with a POS LV, NGS used an Ashtech (Thales Navigation, Santa Clara, Calif.) Z Xtreme GPS base station with RTK capabilities during data collection. The GPS base station was set up over a geodetic control marker for which the best available height value was known. Alternatively, data from a Continuously Operating Reference Station (CORS) could be used, if located within 15 km of the survey area. During the data collection phase, a personal computer was interfaced with the POS LV to monitor the quality of data being collected. Data collection with the POS LV consisted of three steps: pre-survey static initialization, roving data collection and post-survey static initialization.

Pre-survey static initialization consisted of keeping the POS LV-equipped vehicle stationary while in close proximity to the GPS base station (after beginning data collection) for a period of 10 minutes. The initialization period allowed for a static data set to be used during data reduction, and was required for the POS LV to orient itself effectively. Roving data collection consisted of driving the vehicle at highway speed along the selected hurricane evacuation routes. The data collection area was confined to a circle with a 15 km radius with the GPS base station (or CORS site) located at the center. If loss of lock with GPS signals occurred during data collection, the vehicle was stopped as soon as possible in an area conducive to GPS signal reception. At that point, a 10-minute static data set was collected before resuming roving data collection.

As in pre-survey, post-survey static initialization required keeping the vehicle stationary for 10 minutes while in close proximity to the GPS base station. The initialization periods allowed for static data sets to be used during data reduction.

Data Reduction

POS LV data was post-processed using POSPac software. The POS LV data reduction software merged data from all sensors into a single file; differentially processed POS LV/GPS data relative to GPS base station data; blended reduced GPS data with IMU data; and output reduced data at user-defined spacing in an ASCII format. It also allowed for analysis of data for quality control purposes during all stages of data reduction. For example, after blending the IMU data with the post-processed GPS data, overall RMS (Root Mean Square) values were given for the data sets. These RMS plots allowed users to evaluate the overall relative accuracy of the collected data.

Data Validation

While data was being collected with the POS LV, validation points were also being established along the data collection route using an RTK-roving GPS receiver. These validation points were established near the base station and the middle point along the route, and at extreme ends of the data collection area. The height values of these validation points were used as a direct comparison against the height value of the closest available POS LV-generated profile point.

A control network consisting of 80 stations was established simultaneously with POS LV data collection. Forty of these control stations were used as base stations for the POS LV during data collection. At the time of data collection, the best available coordinates for these stations were held as fixed control for POS LV data reduction. Due to the subsidence of the land mass, and thus of the existing control stations in the region, the adjustment and analysis of this control network was complicated and is not yet fully complete (see "Prioritizing the Katrina Evacuation" sidebar on page 26). Once the final coordinates have been determined for these control stations, a correction will be applied to the POS LV base stations and the profile point data based upon the difference between the initial height used during data reduction and the final adjustment height values for these control stations. The relative accuracy of the profile point data will remain consistent despite the eventual change in height values. The absolute height values of the profile point data will change accordingly with the final adjustment values determined for the control network. After the appropriate corrections have been applied to the profile point data, based upon the final adjustment values for the corresponding base stations, profile overlap (data collected from different base stations) will provide for additional analysis of the absolute accuracy of the data collected.

Final Product

The ASCII output file generated by the POSPac software was easily converted into a standard GIS file (e.g. an ESRI Shape file). At this stage, measurement from the road surface to the vehicle reference point will need to be accounted for in the data set. Then the profile data can be graphically displayed and analyzed by users, who include emergency preparedness specialists and GIS professionals in southern Louisiana. They will use the data to analyze roadways and levee tops to determine low spots, or areas most likely to first become inundated, and to make comparisons between the actual elevation and what was thought to be the elevation in a given area.

Moving Forward

Although much of the Gulf Coast has been devastated by Katrina, the data from this project will still be of use. Charlie Challstrom, NGS director said, "We need to recognize the importance of the NGS role in measuring subsidence for making emergency response decisions, but also for the rebuilding efforts that need to come." Once the water and debris is cleared, the roads in southern Louisiana will be used once again. For example, as the massive offshore oil industry in the area re-establishes its presence and functions, it will need to be supplied with goods by trucks that will pass along many of the roads involved in this project. Knowing the accurate elevations of these roads will be of great value to those involved in rebuilding efforts, as well as those involved in preparing for future emergency situations.

Rebuilding the Future

The utter devastation inflicted on the Gulf Coast region by Hurricane Katrina serves as a reminder of our fragile existence and our vulnerability to the forces of nature. As increasing numbers of people move to the coasts of this nation, we face unprecedented levels of risks, in both dollar figures and human lives, from natural disasters such as hurricanes and tsunamis.

In the city of New Orleans, and throughout the Gulf Coast region, people will need to rebuild their lives, their businesses and-in all too many cases-their households. It is in this process of rebuilding that the NGS and land surveyors will play a critical role. If it was not evident before, it should certainly be clear now that accurate elevations are critical to ensure the safety of structures and human lives. In the dynamic coastal areas of the Gulf of Mexico, traditional concrete monuments and "stationary" bench marks will not suffice; establishment of a highly accurate, dynamic survey control network that can cope with changing elevations is vital.

The NOAA Height Modernization Program will be fundamental in the rebuilding of New Orleans and the Gulf Coast region and will usher in a new elevation paradigm. The Louisiana Height Modernization Program is a collaborative effort led by NOAA and the Louisiana Spatial Reference Center at Louisiana State University that seeks to efficiently update and accurately correlate the current GPS and NAVD 88 heights, as well as monitor future changes in elevations using GPS and Continuously Operating Reference Stations (CORS).

The NGS will continue to work on the establishment of GULFNET, a network of CORS located in the Gulf Coast region. By using CORS as the basis for the survey control network, constantly updated positional information, accurate to the sub-centimeter level, will be available throughout the region. Traditional monuments will tie together the GPS (ellipsoid) and NAVD 88 (orthometric) height systems, an important link in modernizing elevation data. The ability to monitor changes in elevations over large areas with unprecedented accuracy and speed will be possible using new technologies such as the one described in the article here. By combining the information from CORS with processing software like the Online Positioning User Service (OPUS), the NGS will enable surveyors and other users of spatial data with the means to determine positions with unprecedented accuracy and speed.
-Casey Brennan

On Aug. 28, 2005, the mayor of New Orleans ordered a mandatory evacuation for the city. Shelters were also set up for those unable to leave, including the Superdome. Photo courtesy of FEMA.

Sidebar: Prioritizing the Katrina Evacuation

The NGS provided preliminary values from this project to three participating agencies in July 2004 after the final data collection sessions had ended-the Lafourche and Jefferson parish levee districts and the Louisiana Department of Transportation. The NGS does not normally provide preliminary data, but made an exception since these partners were directly involved in support roles of the project and this information was of extreme importance to them. The recipients were warned that while the preliminary profile point relative accuracy would remain constant within the individual data sets, the absolute accuracy of the preliminary profile data would change with future adjustments of the data as a whole. In other words, although the preliminary profile data NGS released could be used to determine things such as low areas within the data sets, the data could not be used to determine actual elevations.

Despite the data's limitations, the partners were able to use it prior to Hurricane Katrina's landfall. According to Charlie Challstrom, NGS director, "They [the partners] found that it was extremely helpful for them to prioritize which areas to evacuate first. State and local agencies understood that there was more danger than previously anticipated because of the subsidence." Challstrom added that the NGS data "may have contributed to the overall decision to evacuate earlier."