Mapping the Zone: Improving Flood Map Accuracy (2009)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Summary F loods are the leading cause of natural disaster assess the benefits and costs of more accurate flood losses in the United States, costing approxi- maps, and recommend ways to improve flood mapping, mately $50 billion in property damage in the communication, and management of flood-related data. 1990s alone. To manage flood risk and minimize The charge to the committee is given in Box S.1. future disaster relief costs, the nation invests significant The committee based its findings and recommen- resources in mapping flood hazard areas and providing dations on information gathered from presentations, federal flood insurance to residents in communities that publications, and case studies carried out by the com- regulate future floodplain development. The Federal mittee and the North Carolina Floodplain Mapping Emergency Management Agency’s (FEMA’s) Flood Program, which has high-accuracy data and maps for Insurance Rate Maps (FIRMs, hereafter referred to as nearly the entire state, enabling comparison of new flood maps) are used for setting flood insurance rates, and traditional data and techniques. The case studies regulating floodplain development, and communicat- focused on (1) uncertainties in hydrologic, hydraulic, ing the 1 percent annual chance flood hazard to those and topographic data in and near selected streams in who live in floodplains. Florida and North Carolina, and (2) the economic costs Making and maintaining an accurate flood map is and benefits of creating new digital flood maps in North neither simple nor inexpensive. FEMA’s Map Modern- Carolina. The North Carolina analyses were carried out ization Program, funded for fiscal years 2003 to 2008, in three physiographically distinct areas: mountains will result in flood maps in digital format for 92 percent (city of Asheville), rolling hills ­(Mecklenburg County), of the continental U.S. population. Taking flood maps and coastal plain (Pasquotank and Hertford Counties). into the digital world was a great step forward because For the economic analysis, two benefits were consid- digital maps are more versatile for floodplain manage- ered, based in part on the availability of geospatial data ment and other uses and they are easier to update. Yet required to carry out the analysis: (1) avoiding flood even after an investment of more than $1 billion, only losses to new buildings and avoiding repairs to infra- 21 percent of the population has maps that meet or structure through accurate floodplain delineation, and exceed national flood hazard data quality thresholds (2) setting flood insurance premiums to better match (Figure S.1). Even when floodplains are mapped with estimates of actual risk. high accuracy, land development and natural changes to the landscape or hydrologic systems create the need FACTORS THAT AFFECT for continuous map maintenance and updates. FLOOD MAP ACCURACY FEMA and the National Oceanic and Atmo- spheric Administration (NOAA) sponsored this study The components of FEMA flood maps that are to examine the factors that affect flood map accuracy, most relevant to the issues of accuracy discussed in this  

 MAPPING THE ZONE Status of National Flood Hazard Data Quality Standards as of March 31, 2008 C A N A D A uperio e S r ak L La ke n Hu La ke Michiga o r on tari e On Lak ie Er ke La A t l a n t i c O c e a n P a c i f i c O c e a n Alaska RUSSIA M E X I C O CANADA Hawaii LEGEND Meets or Exceeds National Flood Kauai Hazard Data Quality Thresholds Niihau Molokai Oahu Approaches National Flood Maui G u l f o f Hazard Data Quality Thresholds Lanai Pacific Pacific Digital Product Issued M e x i c o Ocean Ocean Hawaii Modernized Map Not Yet Issued 0 500 1,000 Miles 0 100 200 Miles Study Not Planned 0 125 250 500 750 1,000 Miles Projection: North America Albers Equal-Area Conic FIGURE S.1  Data quality standards achieved by individual counties as of March 31, 2008. Green counties meet or exceed national Figure S-1.eps flood hazard data quality thresholds. Yellow counties meet some standards. In red counties, the maps have been updated digitally and a digital product has been issued. Compliance with data quality standards was not required for such digital conversions, although a limited FEMA audit suggests that some portions of these counties meet the standards. In beige counties, modernized maps have not yet been issued because the first phase of map production has not been completed or quality data do not exist. No study is planned in white counties. SOURCE: Paul Rooney, FEMA. report are the floodplain boundaries and base flood ele- delineate the predicted floodplain area. The process vations. Floodplains are low-lying, relatively flat areas is similar for coastal flood mapping, except the exist- adjoining inland and coastal waters. The most common ing repository of observational data (hurricane winds, floodplains mapped are those created by the 1 percent topography, and bathymetry) is smaller and extreme annual chance flood (also known as the 100-year flood) events are more difficult to capture. As a result, coastal and the 0.2 percent annual chance flood (also known flood maps rely more heavily on modeling of wave and as the 500-year flood). The base flood elevation is the erosion processes and storm surge (water that is pushed computed elevation to which floodwater is expected to toward the shore by the force of winds swirling around rise or that it is expected to exceed during a 1 percent a storm) to predict coastal flood elevations. All of the annual chance flood, and it forms the basis for set- inputs have uncertainties that affect the accuracy of the ting flood insurance premiums and structure elevation resulting flood map. regulations. The extent of potential flood inundation must be OVERARCHING FINDINGS predicted from statistical analyses and models. For riv- erine flooding, statistical estimates of flood discharges Finding 1. Topographic data are the most important at U.S. Geological Survey (USGS) stream gages and factor in determining water surface elevations, base digital representations of the land surface topography flood elevation, and the extent of flooding and, thus, provide data for hydrologic and hydraulic models. The the accuracy of flood maps in riverine areas. output is used in geographic information systems to 

SUMMARY  BOX S.1  Committee Charge The committee will 1. Examine the current methods of constructing FEMA flood maps and the relationship between the methods used to conduct a flood map study (detailed study, limited detailed study, automated approximate analysis, or redelineation of existing hazard information), the accuracy of the predicted flood elevations, and the accuracy of predicted flood inundation boundaries. 2. Examine the economic impacts of inaccuracies in the flood elevations and floodplain delineations in relation to the risk class of the area being mapped (based on the value of development and number of inhabitants in the risk zone). 3. Investigate the impact that various study components (i.e., variables) have on the mapping of flood inundation boundaries: a. Riverine flooding • The accuracy of digital terrain information • Hydrologic uncertainties in determining the flood discharge • Hydraulic uncertainties in converting the discharge into a floodwater surface elevation b. Coastal flooding • The accuracy of the digital terrain information • Uncertainties in the analysis of the coastal flood elevations c. Interconnected ponds (e.g., Florida) • The accuracy of the digital terrain information • Uncertainties in the analysis of flood elevations 4. Provide recommendations for cost-effective improvements to FEMA’s flood study and mapping methods. 5. Provide recommendations as to how the accuracy of FEMA flood maps can be better quantified and communicated. 6. Provide recommendations on how to better manage the geospatial data produced by FEMA flood map studies and integrate these data with other national hydrologic information systems. A study of sampling uncertainties in extreme stage 0.9 foot). They are least pronounced in mountainous heights at USGS stream gages in North Carolina areas, raising the base flood elevation an average of and Florida found that for 30 of 31 gages, the aver- 0.2 foot, which is not significant, given the sampling age uncertainty is approximately 1 foot with a range uncertainty noted above. of 0.3 feet to 2.4 feet. Uncertainties do not appear to The largest effect by far on the accuracy of the base vary with the size of the drainage basin or its topo- flood elevation is the accuracy of the topographic data. graphic slope. It may thus be inferred that the lower The USGS National Elevation Dataset (NED), devel- bound on the uncertainty of the base flood elevation oped from airborne and land surveys, is commonly used is approximately 1 foot. For the river reaches studied in flood map production, even though the elevation in North Carolina, a 1-foot change in flood elevation uncertainties of the NED are about 10 times greater corresponds to a horizontal uncertainty in the flood- than those defined by FEMA as acceptable for flood- plain boundary of 8 feet in the mountains, 10 feet in plain mapping. Data collected using high-resolution the rolling hills, and 40 feet in the coastal plain. This remote sensing methods such as lidar (light detection uncertainty has a significant impact on the delineation and ranging) can have absolute errors on the order of of inundated areas on flood maps. centimeters, consistent with FEMA requirements, but The constriction of flood flow by bridges and they are not available nationwide. A comparison of culverts raises the base flood elevation in the three lidar data and the NED around three North Carolina study areas. Such backwater effects are largest just streams revealed random and sometimes systematic upstream of the constriction and diminish progres- differences in ground elevation of about 12 feet, which sively upstream. They are most pronounced in the significantly affects predictions of the extent of flood- coastal plain, extending an average of 1.1 miles and ing (e.g., Figure S.2). These large differences exceed raising base flood elevations by up to 2.5 feet (average FEMA’s stated error tolerances for terrain data by an 

 MAPPING THE ZONE FIGURE S.2  Inundation maps of the area where the Tar-Pamlico River empties into Pamlico Sound of North Carolina. The figure on the left is based on a digital elevation model (DEM) with 30-meter post spacing created from the USGS NED. The figure on the right is Figure 3-13.eps based on a DEM with 3-meter post spacing created from North Carolina Floodplain Mapping Program lidar data. The dark blue tint represents land that would become inundated with 1 foot of storm surge or sea level rise. The light blue area represents uncertainty bitmap image in the extent of inundation at the 95 percent confidence level. SOURCE: Gesch (2009). order of magnitude and support the need for new topo- for calculating wave heights (Wave Height Analysis graphic surveys, as called for in a National Research for Flood Insurance Studies [WHAFIS]), which was Council (NRC, 2007) report Elevation Data for Flood- introduced in the late 1970s, with a two-dimensional plain Mapping. In two of the study areas, random errors wave model would improve the accuracy of calculated in topographic data produce inaccuracies in floodplain base flood elevations. Coupled two-dimensional surge boundaries, but do not significantly alter the total area and wave models, as well as models that account for ero- of the floodplain. In the other study area, in addition sion processes, the effects of structures, and variations to random errors, there is a large systematic difference in topography, offer the potential for further improve- between the lidar and NED data that results from a ments of coastal flood map accuracy. A comparison of misalignment of the stream location between the base available models, conducted by an independent external map planimetric information and the topographic data. advisory group, would help quantify uncertainties and As a result, the total areas of the floodplains defined indicate which models should be incorporated into from lidar and from the NED differ by 20 percent. mapping practice. Because imagery is improving faster than elevation, the misalignment problem is growing more acute. Finding 3. Flood maps with base flood elevations yield greater net benefits than flood maps without. Finding 2. Coastal flood maps can be improved sig- nificantly through use of coupled two-dimensional Benefit-cost analyses have shown that the greatest storm surge and wave models and improved process benefits of more accurate flood maps are avoided flood models, which would yield more accurate base flood losses to planned new buildings and avoided repairs elevations. to infrastructure through more accurate base flood elevations and depiction of floodplain boundaries. The science of riverine flooding is reasonably well Producing a more accurate base flood elevation yields understood, and improvements to inland flood maps can the greatest increment of benefits because it enables focus on harnessing available technology. In contrast, insurance premiums and building restrictions to be set advancing understanding of the complex ­dynamics of commensurate with a more realistic profile of the hori- the coastal inundation process is necessary for improv- zontal and vertical extent of flooding. Only the more ing the accuracy of coastal flood maps. Coastal flood expensive of FEMA’s flood study methods—detailed models are evolving rapidly, but published results sug- studies and most limited detailed studies—yield a base gest that replacing FEMA’s one-dimensional model flood elevation. A comparison of study methods in the 

SUMMARY  three case study areas by the North Carolina Floodplain located, but also the likely consequences of flooding Mapping Program showed that the use of detailed (e.g., damage to houses, coastal erosion). Inundation studies and limited detailed studies that generate base and risk maps beginning to be produced by U.S. federal flood elevations results in net benefits to the state. In and state government agencies and by other countries contrast, the use of approximate study methods, which have attributes that merit FEMA’s attention. do not yield base flood elevations, results in net costs. Maps that show only floodplain boundaries have This is significant because detailed and limited detailed the disadvantage of implying that every building in studies in North Carolina rely on lidar data, and even a designated flood zone may flood and that every though lidar surveys are expensive, the costs to map building outside the zone is safe. Providing floodplain the three study areas are outweighed by the benefits of residents with the elevation of structures relative to the more accurate maps. expected height of a number of floods offers a better way to define graduated risk (from low risk to high Finding 4. The most appropriate flood study method risk). Where the necessary data are available (e.g., to be used for a particular map depends on the accu- structure elevation, base flood elevations, flood protec- racy of the topographic data and the overall flood risk, tion structure performance), a geographic information including flood probability, defined vulnerabilities, system could be used to personalize flood risk to indi- and consequences. vidual addresses. The North Carolina benefit-cost analysis showed RECOMMENDATIONS that a combination of different study methods produces the greatest economic benefits to the state as a whole. The body of the report contains focused recom- The best study method depends on the characteristics mendations on how to improve specific aspects of of the area being mapped, such as the present and future FEMA’s flood data, models, and mapping. The follow­ potential of flooding, the potential for population ing overarching recommendations address Tasks 4 growth, the availability of land for development, and through 6 and are based on the analysis of information the likely economic value of structures to be built. The presented throughout the report. quality of the topographic data is also important. Where accurate topographic data are available, an ­accurate base Cost-Effective Improvements to FEMA’s Flood flood elevation can be calculated, a more accurate map Study and Mapping Methods can be produced, and thus better decisions can be made about appropriate use of the floodplain. Recommendation 1. FEMA should increase collabo- ration with federal (e.g., USGS, NOAA, U.S. Army Finding 5. FEMA’s transition to digital flood map- Corps of Engineers), state, and local government ping during the Map Modernization Program ­creates agencies to acquire high-resolution, high-accuracy opportunities for significant improvements in the topographic and bathymetric data throughout the communication of flood hazards and flood risks nation. through maps and web-based products. Riverine mapping methods are well established, FEMA is moving from simply portraying flood although improvements could be made in calibrating hazard and flood insurance rate zones on maps to rainfall-runoff models, updating regression equations communicating and assessing risk, an ambitious goal (many of which are more than 10 years old) more that leverages the digital flood-related information frequently, and increasing the use of two-dimensional and maps produced during the Map Modernization models developed by the research community. The Program as well as FEMA tools for estimating flood greatest improvement, however, would come from use damage and loss (i.e., Hazards U.S. Multi-Hazards of high-accuracy, high-resolution topographic data. software). To communicate risk, the maps and prod- Improved measurements of channel, lake, estuarine, ucts must show not only where flood hazard areas are and near coastal bathymetry would augment the 

 MAPPING THE ZONE improved measurement of land surface topography Quantifying and Communicating the Accuracy of enabled by lidar technology. As noted above, the use of FEMA Flood Maps lidar data to calculate more accurate base flood eleva- tions and floodplain boundaries reduces future flood Recommendation 4. FEMA should require that every losses and produces net benefits to the State of North flood study be accompanied by detailed metadata Carolina. Reducing future flood losses also benefits identifying how each stream and coastline reach taxpayers throughout the nation. FEMA has recently was studied and what methods were used to identify begun to support collection of lidar data along the Gulf the magnitude and extent of the flood hazard and to coast, but lidar data coverage over most inland areas is produce the map. still sparse. One of the most important ways to quantify and Recommendation 2. FEMA should work toward communicate flood map accuracy is to document a capability to use coupled surge-wave-structure the data and methods used to study each segment of m ­ odels to calculate base flood elevations, starting stream or coastline. FEMA’s current metadata report- with incorporating coupled two-dimensional surge ing requirements do not include all the information and wave models into mapping practice. needed to assess the quality and reliability of the data under­lying the maps. For each stream or coastline mile A significant improvement to coastal flood map- studied, metadata should describe what input data, ping can be made by improving the models. Currently, mapping, and modeling methods were used; the date base flood elevations are calculated by combining of mapping; the contractor; and the starting and end- storm surge models with wave models, and using the ing points. result in models that calculate erosion and wave effects. However, modeling has greatly advanced, and it is Managing Geospatial Data now possible to use coupled models that account for storm surge, waves, erosion, and topographic features Recommendation 5. FEMA should reference all simultaneously. stream and coastal studies within its Mapping Infor- mation Platform to the USGS National ­Hydrography Recommendation 3. FEMA should commission a Dataset. scientific review of the hydrology and hydraulics needed to produce guidelines for flood mapping in FEMA Map Modernization has produced a large ponded landscapes. amount of geospatial data and flood hydraulic models for the nation’s streams and coastlines. The result is Methods to map landscapes in which water tends the most comprehensive digital description of the to flow from one ponded area to the next (shallow nation’s streams and rivers that has ever been under- flooding) are still being developed. The primary hurdle taken. These data are stored in the Mapping Informa- to progress is the lack of scientific studies and models tion Platform (MIP) on a county-by-county basis. on the interactions between ponds, the volume of water There is no requirement that map information such temporarily stored in the depressions, and the rate at as stream centerlines be consistent from one county to which it percolates out. Commissioning a study would the next. The USGS National Hydrography Dataset not be costly and is a necessary step toward improving is a seamless, connected map of the nation’s streams, shallow flood mapping. rivers, and coastlines. Using a technique called linear referencing, it is feasible to link the FEMA stream and coastline data with the corresponding information in the National Hydrography Dataset. If this were done, FEMA flood data could become an integral part of the nation’s hydrologic information infrastructure rather than existing as a separate database.