The nation is presently making an investment of more than $200 million per year in Flood Map Modernization. The goal is to replace paper Flood Insurance Rate Maps (FIRMs) by Digital Flood Insurance Rate Maps (DFIRMs), and to achieve national coverage with these digital maps. Resource limitations have meant that the scope of the effort has been modified to increase map quality by mapping 65 percent of the nation that contains 92 percent of its population. This work is being carried out by the Federal Emergency Management Agency (FEMA) in collaboration with state and local partners as part of the National Flood Insurance Program (NFIP). This program encourages communities to regulate the land development in their floodplains to avoid flood damages and, in return, allows property owners located in flood hazard areas to purchase federal flood insurance. This insurance is designed to provide an alternative to federal disaster assistance to reduce the costs of repairing damage to buildings and their contents caused by floods. Federal flood insurance is required if the property owner has a federally-backed mortgage.
During the annual appropriations hearings for Flood Map Modernization, concerns have been expressed to Congress that the underlying framework data used as input to the flood mapping process is not of adequate quality in much of the nation to properly support the new digital flood map creation. This study was commissioned by the National Academies for the purpose of informing Congress and the nation about this issue.
The National Academies requested that an ad hoc committee be established to respond to the following statement of task:
Identify the current mapping technologies being used by FEMA to develop flood hazard maps;
Identify mapping technologies that are currently available; and
Determine if newer technologies are appropriate and would be of additional benefit to floodplain mapping.
This study was conducted in a short period of time to enable Congress to consider its conclusions during deliberations on appropriations in the spring of 2007. Limitations of time, and the narrow focus of the statement of task, meant that this study did not focus in detail on the following issues: (1) coastal flooding—this involves a different methodology than riverine flooding and since the nation has 60,000 miles of coastlines and about 4.2 million miles of rivers and streams, the committee focused on riverine flooding because that makes up the bulk of Flood Map Modernization; (2) geodetic control—the committee did not consider variations in the precision of definition of the survey control points and
vertical datums across the nation, but this report does highlight land subsidence as an issue important to Flood Map Modernization; (3) mapping technologies other than airborne remote sensing—the committee considered in detail only photogrammetry, light detection and ranging (lidar), and interferometric synthetic aperture radar (IFSAR), which are all aerial mapping technologies, and did not consider land-based mapping technologies; (4) uncertainties in flood hydrology and hydraulics—the committee focused on issues related to base map and elevation data and not on hydrologic and hydraulic sources of uncertainty in flood risk assessment. A previous study examined uncertainty in flood hydrologic and hydraulic computations (NRC, 2000).
Besides the mapping technologies study presented in this report, FEMA has separately engaged the National Academies to undertake a longer-term flood map accuracy study within which the above issues will be addressed more fully.
This report presents the committee’s response to its statement of task. Two aspects of mapping need to be considered in this context—defining land surface reference information and land surface elevation.
BASE MAP INFORMATION
Land surface reference information describes streams, roads, buildings, and administrative boundaries that show the background context for mapping the flood hazard zone. The older paper FIRMs contain only vector data (points, lines, and polygons) to describe all land surface reference features. The new DFIRMs typically use a digital orthophoto as the base map, supplemented by planimetric vector data for key map features (e.g., roads needed for georeferencing building locations) and administrative boundaries (e.g., city or county boundaries) that cannot be observed in photography. An orthophoto is an aerial photograph from which all relief displacement and camera tilt effects have been removed such that the scale of the photograph is uniform and it can be considered equivalent to a map. FEMA’s requirement for an orthophoto base map is that it meet or exceed the U.S. Geological Survey (USGS) Digital Orthophoto Quarter Quadrangle (DOQQ) specification, which calls for a 1-meter pixel resolution, or ground sample distance (GSD), and the meeting of National Map Accuracy Standards at a scale of 1:12,000. The popularity of Google Earth (http://earth.google.com) has introduced into the public mind the idea of “image as base map” and reinforces the importance of regularly updated digital orthophotography covering the nation. The committee believes this base image mapping standard is satisfactory and the nation has adequate image mapping to support Flood Map Modernization through the National Digital Orthophoto (http://www.ndop.gov) and National Agriculture Imagery (http://www.fsa.usda.gov/FSA/apfoapp?area=home&subject=prog&topic=nai/) programs. The committee endorses the proposed Imagery for the Nation program, which seeks to create and maintain 1-meter GSD or better orthophoto products seamlessly across the United States
(http://www.nsgic.org/hottopics/imageryofnation.cfm). Because the committee concludes that the nation’s existing base mapping for land surface reference information is adequate, this report concentrates on the elevation data input to floodplain mapping, which has a much greater effect on the accuracy of floodplain maps and an important component of those maps, the Base Flood Elevation (BFE).
BASE FLOOD ELEVATION
Land surface elevation information defines the shape of the land surface, which is important in defining the direction, velocity, and depth of flood flows. Land surface elevation data for flood management studies of individual streams and rivers have traditionally been derived by land surveying, but the very large areal extent of FEMA floodplain mapping, which covers nearly 1 million miles of the nation’s streams and shorelines, means that land surface elevation data for Flood Map Modernization are mostly derived from mapped sources, not from land surveying. Land surface elevation information is combined with data from flood hydrology and hydraulic simulation models, to define the BFE, which is the water surface elevation that would result from a flood having a 1 percent chance of being equaled or exceeded in any year at the mapped location. A floodplain map is created by tracing the extent of inundation of the landscape by water at the BFE.
The insurance industry uses floodplain maps to determine if lenders require purchasers of new buildings to have federal flood insurance. This insurance is required if any part of the footprint (or plan view) of the building outline lies over the spatial extent of the floodplain. In other words, the flood insurance determination is made on the basis of a planimetric or horizontal criterion: Does the building outline lie within the floodplain? The current FIRMs and DFIRMs properly support this flood insurance process.
Use of the maps to regulate land development in floodplains by local communities typically requires the first floor elevation of a building to be at or above the BFE if that building is to be constructed within the floodplain. The governing criterion used is thus often stated as: Is the first floor elevation above the BFE? In some communities, a safety margin such as 1 foot of elevation is added to the BFE in order to take into account allowable encroachments into the floodplain that may raise the water surface elevation by 1 foot. This criterion, based on vertical rather than horizontal criteria, is better than that used in flood insurance determinations.
Rational floodplain management and flood damage estimation depend not only on how far the water spreads, but also on how deeply buildings are flooded and with what frequency. If the task of the nation’s flood management is observed in this larger context, accurate land surface and floodwater surface elevation information are critical. For example, in the flood damage mitigation projects undertaken by the United States Army Corps of Engineers in collaboration with local communities, flood damage estimation requires knowing the first
floor elevation of all flood-prone buildings. FEMA also requires that the flood depth at structures be known for detailed study areas when flood insurance is obtained. The flood insurance rate for detailed study areas is based on the height of the first finished floor with respect to the BFE. The committee concludes that rational flood management for the nation requires that the problem be viewed in three dimensions, quantifying flood depth throughout the floodplain, not as a two-dimensional problem of defining the extent of a floodplain boundary on a flat map.
If a property owner whose building is classified for insurance purposes as being within the floodplain wishes to protest that determination, a laborious and expensive procedure is undertaken, for both the owner and the government, to process a Letter of Map Amendment (LOMA). About 15,000 LOMAs are currently being processed per year, and the work and expense involved probably prevent many more owners from getting a LOMA. This facet of its implementation makes the NFIP more of a burden on individual citizens than it could be. Even when an owner obtains a LOMA to avoid purchasing flood insurance, the property is still at risk, perhaps just below the 1 percent annual chance flood. In these cases the owner can still obtain flood insurance, but the flood policy is at a much reduced rate.
The committee concludes that in order to fully support the NFIP, updated floodplain maps should show the BFE as well as the spatial extent of the floodplain boundary. Displaying these features requires high-accuracy elevation data.
So far, updated DFIRMs have been prepared for about 1 million of the nation’s 4.2 million stream miles. Of the approximately 1 million stream miles completed up to June 2005, one-quarter (or 247,000 stream miles) were mapped using detailed studies employing high-resolution elevation data, and the resulting flood maps show the BFE of the floodwater surface as well as the spatial extent of the floodplain. The remaining three-quarters (or 745,000 stream miles) were frequently mapped with more approximate engineering methods, and the resulting maps show only the spatial extent of the floodplain but not its BFE.
FEMA Map Modernization requires elevation data for floodplain mapping to represent the current conditions in the area, or to be supplemented with updated information. The existing National Elevation Dataset (NED)1 results mainly from the interpolation to a grid cell format of the elevation contours depicted in standard USGS 1:24,000-scale topographic maps. These maps were made over a long period with a peak emphasis during the 1960s and 1970s, such that, on average, the date of origin of these maps is 1970. Some new high-
The USGS National Elevation Dataset has been developed by merging digital elevation data available across the United States into a seamless raster format. The dataset provides coverage of the United States, Hawaii, Alaska, and the island territories. The data are available for free download at http://ned.usgs.gov.
resolution elevation data have been added to the NED in local areas, but for the most part, the nation’s description of its land surface elevation is more than 35 years old. A great deal of land development and urban expansion has occurred since, which has materially altered the shape of the land surface. In particular, the existence of new road embankments and flood drainage structures can significantly impact floodwater surface elevations, especially in flat areas. Existing elevation data in these areas are frequently out of date.
Land subsidence affects 17,000 square miles of land area in 45 states, especially California, Texas, Louisiana, and Florida. The land surface in these areas is sinking at a significant rate, which has rendered old elevation data obsolete. This is particularly significant in coastal areas because subsidence of coastal lands leads to greater inundation from the sea.
FEMA requires elevation data of 2-foot equivalent contour accuracy by National Map Accuracy Standards in flat areas and 4-foot equivalent contour accuracy in rolling and hilly areas, which corresponds to a root mean square error of 0.61 feet (18.5 centimeters) for flat areas and 1.22 feet (37.0 centimeters) for rolling and hilly areas, respectively. Flat and hilly are not defined quantitatively in the current FEMA guidelines; they are subjective terms that are to be interpreted during the scoping phase of a flood study. The existing NED has a root mean square error when compared to National Geodetic Survey control points of 7.68 feet (2.34 meters). In other words, the FEMA detailed floodplain mapping standards call for elevation data that are about 10 times more accurate than the NED, although existing elevation data coverage in many areas of the country is of significantly better quality. This means that the existing NED, and the topographic contour information upon which it is based, are not adequate to support Flood Map Modernization, except where new high-accuracy elevation data have been added from state or local sources.
However, it should be noted that the NED is a very effective means of combining elevation data from many sources and serving them to the public in a seamless way, and the committee is supportive of its continuing mission to achieve this goal of public access to the nation’s elevation data. The issue is with the age and inaccuracy of most of the information in that dataset.
It is shown in this report that in the existing NED, 11 percent of the land area of the continental United States and of Alaska is determined to have zero slope. These areas are located along the Gulf coast, in Florida, along the Eastern seaboard, and at various locations in the interior of the nation. High-accuracy elevation data are especially needed in these areas to support floodplain mapping.
Some communities undertaking Flood Map Modernization already have available elevation data of the required accuracy or pay to acquire such data as part of their contribution to the costs of floodplain mapping. Apart from exceptional circumstances, FEMA does not pay for the costs of elevation data acquisition in local communities. This means that for many communities, the NED and the associated 1:24,000 topographic contours are
the best elevation information available for floodplain mapping. The committee concludes that the elevation data in the existing NED are too old and inaccurate to support FEMA Map Modernization.
ELEVATION FOR THE NATION
A new measurement program for the land surface elevation is needed, which the committee has termed Elevation for the Nation. At present, individual communities and some states are undertaking such elevation measurement programs over part or all of their jurisdictions, but there is no guarantee that this uncoordinated process will produce the required elevation data consistency and accuracy. The committee has concluded that elevation data of 1-foot equivalent contour accuracy are required in very flat coastal or inland floodplains for the whole country as part of a national elevation program.
At the outset of the Flood Map Modernization program, new laser- and radar-based elevation measurement technologies were emerging. However, they had not been widely adopted commercially and their costs were so high that to remap the elevation of the entire nation was considered prohibitively expensive. Since that time, light detection and ranging (lidar) has matured to become what this committee concludes is the preferred technology for elevation mapping. Also, mapping costs have fallen as the technology has been more widely adopted by community and state mapping programs, and the use of lidar to measure land elevation quickly and accurately has become a preferred practice.
Lidar operates by projecting short laser pulses of light from an aircraft and measuring the time taken for these pulses to bounce back to the aircraft from the land surface. With appropriate processing, 1-foot to 2-foot equivalent contour accuracy can be achieved in the final bare-earth elevation data; this level of accuracy meets or exceeds FEMA elevation criteria for floodplain mapping in all areas. During the committee’s public session, presentations were made by representatives of a number of federal agencies, and the committee was struck by the degree of agreement among them that lidar is now the technology of choice for land surface elevation measurement.
Lidar data acquisition from aircraft produces a dense cloud of measured points, some of which define the land surface while others reflect off vegetation and trees above the ground. This requires processing the raw measured data to extract those points representing the bare-earth elevation. Lidar pulses do not reflect off water in the same way they reflect off land; smooth surfaces reflect very few pulses back to the aircraft and thus often appear as void areas in the dataset. Furthermore, the presence of overhanging trees near streams makes locating banklines of rivers and shorelines of water bodies difficult from the lidar points alone. Additional interpretation of lidar data and photogrammetry is needed to define breaklines correctly in the landscape representing banklines, shorelines, and coastlines separating land and water features.
Elevation for the Nation implies not simply a new data measurement initiative, but also a change in the way the nation’s elevation data are archived. In order to support all forms of subsequent interpretation, all of the measured lidar points should be stored. The points defining the bare-earth elevation are combined with breaklines defining the boundaries of water features to produce a digital terrain model that is capable of several forms of output representation, including traditional contours, regularly gridded digital elevation models, or a better approach called a triangulated irregular network (TIN), in which individual points and lines are combined into a triangular mesh that continuously spans the land surface of an area. TINs represent sharp land surface features such as road embankments precisely, and they are the representation of choice for hydraulic analysis of floodplains, which defines floodwater surface elevations.
A number of states and local communities are acquiring new elevation data on their own initiative and for various purposes, but these datasets frequently do not satisfy FEMA guidelines and specifications, for example, satisfying 10-foot equivalent contour accuracy rather than the 2- to 4-foot equivalent contour accuracy required by FEMA. The committee does not believe that allowing this ad hoc process to continue will create consistent elevation data of the required accuracy to fully support floodplain mapping over the nation. The elevation data collection program undertaken by North Carolina to support its statewide floodplain mapping is highlighted in this report as an example of a state data collection program with the data standards and collection procedures appropriate for a national program. The new high-accuracy elevation data collected in North Carolina are valuable for many other functions in the state.
CONCLUSIONS AND RECOMMENDATIONS
The committee concludes that the nation’s information for land surface elevation is inadequate to support FEMA’s Map Modernization and that new national digital elevation data collection is required. The committee proposes that this program be called Elevation for the Nation to parallel the existing Imagery for the Nation concept. The committee recommends the following:
Elevation for the Nation should employ lidar as the primary technology for digital elevation data acquisition. Lidar is capable of producing a bare-earth elevation model with 2-foot equivalent contour accuracy in most terrain and land cover types; a 4-foot equivalent contour accuracy is more cost-effective in mountainous terrain, and a 1-foot equivalent contour accuracy can be achieved in very flat coastal or inland floodplains. A seamless nationwide elevation database created at these accuracies would meet FEMA’s published requirements for floodplain mapping for the nation. The first focus of this program should be on remapping the elevation of
the 65 percent of the nation that contains 92 percent of its population, where flood risk justifies the required data collection. The program can use newly acquired data or existing local and regional data if the existing data are reasonably up-to-date.
A seamless nationwide elevation model has application beyond the FEMA Map Modernization program; some local and state governments are acquiring lidar data at these accuracies or better. For example, in 2007, the Florida Division of Emergency Management will be acquiring lidar data satisfying 1-foot equivalent contour accuracy of shorelines for storm surge modeling and hurricane evacuation planning. As part of Elevation for the Nation, federal, state, and local mapping partners should have the option to request data that exceed minimum specifications if they pay the additional cost of data collection and processing required to achieve higher accuracies.
The new data collected in Elevation for the Nation should be disseminated to the public as part of an updated National Elevation Dataset.
The Elevation for the Nation database should contain the original lidar mass points and edited bare-earth surface, as well as any breaklines required to define essential linear features.
In addition to the elements proposed for the national database, secondary products including triangulated irregular networks, hydrologically corrected digital elevation models, and hydrologically corrected stream networks and shorelines should be created to support FEMA floodplain mapping. Standards and interchange formats for these secondary products do not currently exist and should be developed. Comprehensive standards for lidar data collection and processing are also needed. Professional societies and federal agency consortia are appropriate entities to lead development of these standards; funding to support these efforts should be considered as part of a nationwide effort.
The committee reached its conclusion that Elevation for the Nation is needed for two main reasons: first, for the nation as a whole the existing elevation data are so old, and the gap between their accuracy and the accuracy required for floodplain mapping is so great, that the need for new elevation data is clear; and second, the required elevation mapping technology exists and has been commercially deployed such that implementing Elevation for the Nation is technically feasible. Regardless of whether “best-available” elevation data are used or new elevation data are being acquired for a flood study, informed judgments must be made about the appropriateness of these datasets and their influence on flood data computations. The committee recognizes that Elevation for the Nation will involve significant expense, perhaps as much as the existing Flood Map Modernization program. It is for Congress and others to determine whether this expense is justified in the context
of national spending priorities. Certainly the data arising from Elevation for the Nation will have many beneficial uses beyond floodplain mapping and management.
The current study was conducted in a short time to address very specific questions about the mapping technologies used to produce floodplain maps. As such, the committee did not have the resources or scope to examine in detail many important issues related to flood map accuracy. The committee suggests, for example, that analysis of a selection of updated flood maps could be useful to compare the quantitative effects of using lidar versus using conventional 10-meter or 30-meter NED information derived from USGS topographic maps to provide the elevation data. In a new, two-year study, beginning in early 2007, FEMA has separately requested the National Academies to undertake a distinct evaluation of flood map accuracy, including an examination of the whole range of uncertainty in flood mapping arising from uncertainty in flood hydrology and hydraulic modeling, as well as uncertainty in land surface elevation. The committee hopes that the present report provides solid input to the upcoming study and helps to further objective examination of the most cost-effective methods needed to support the nation’s floodplain mapping and management.