| Flood Mapping Specifications: Technology-centric vs
Application-centric
By Jeff Allen and Ronald J. Birk IFSAR and LIDAR
During the ASPRS conference in Washington D.C., the week of May 22-26, 2000, there was a notable emphasis on the IFSAR (interferometric synthetic aperture radar) and LIDAR (light detection and ranging) technologies and solutions. No fewer than five educational sessions, a workshop, and an open forum addressed IFSAR and LIDAR technologies, applications, and the development of specifications for the use of IFSAR and LIDAR. One of the important events held during the ASPRS conference was the introduction of the National Digital Elevation Program (NDEP). The NDEP is a consortium which includes, or plans to include, representation of numerous U.S. federal agencies including USGS (United States Geological Survey), FEMA (Federal Emergency Management Agency), NOAA (National Oceanic and Atmospheric Adimistration), USACE (U.S. Army Corps of Engineers), NASA (National Aeronautics and Space Administration), NIMA (National Imagery and Mapping Agency), BLM (Bureau of Land Management), NOS (National Oceanic Service), NGS (National Geodetic Service), TVA (Tennessee Valley Authority), and USFS (U.S. Forest Service). The charter for this program is in the process of being signed by each of these agencies. The NDEP and associated venues at the ASPRS conference addressed a range of topics including floodplain mapping and digital elevation model data collection. The level of interest and energy associated with these sessions was high and is a measure of the importance of these technologies to serve operational requirements for national mapping.
The technologies of IFSAR and LIDAR are proving to have matured to the point of being operational in serving many national mapping needs. IFSAR mapping, provided by Intermap Technologies Inc., has been employed to support projects for NIMA, NASA, and other customers to map locations around the world. LIDAR solutions are being offered by TerraPoint LLC, EagleScan, EarthData, Laser Mapping Specialists, Waggoner Engineering, and many other qualified vendors. Distinction between "Watershed-scale" mapping and "Floodplain-scale" mapping
Development of performance specifications for baseline mapping products is a key issue in enabling government procurements to solicit IFSAR and LIDAR technologies. One important application is related to elevation mapping for Flood Risk Insurance Maps (FIRMs). This application requires two distinct mapping scales, one for the entire watershed, and one for the floodplain areas within a given watershed.
Within the context of this discussion, the extent of watershed areas to be mapped is defined by large areas associated with run-off, drainage, and absorption in (and around) hydrographic features. The boundaries for floodplain areas within the watershed areas are defined as more localized land areas, susceptible to being inundated by floodwaters from any source. Current efforts related to establishing nationally recognized specifications for flood-related mapping are focused on floodplain mapping. There are plans to develop a separate set of specifications for watershed mapping in the near future. Technology-centric versus Application-centric Specifications
At the ASPRS conference, during the opening presentation in the educational session titled "IFSAR and LIDAR: Our Future," David Maune, of Dewberry and Davis, discussed various challenges and applications associated with using IFSAR and LIDAR technologies for flood mapping applications. He recognized the distinction between using LIDAR mapping for the floodplain-scale mapping studies, and the value of using IFSAR mapping for watershed-scale mapping studies. In the second presentation for this session, Jeff Allen, of Michael J. Baker, Jr. Inc., reviewed the status of the Federal Emergency Management Agency (FEMA) in the development of guidelines and specifications for the use of IFSAR and LIDAR for floodplain mapping.
Mr. Allen's presentation described the two new draft FEMA specifications: one draft specification for LIDAR technology to produce floodplain map products, and another draft specifications for IFSAR technology to produce floodplain map products. These specifications provide guidance and requirements of both IFSAR and LIDAR mapping to gather precision elevation data needed to create digital elevation models (DEMs), digital terrain maps, and other National Flood Insurance Program (NFIP) products. The specifications appear as appendices to a document known as FEMA 37, which contains FEMA specifications for photogrammetric solutions for floodplain mapping and related NFIP studies.
A member of the audience questioned the necessity for having three separate specifications for the same map products-given that they were only differentiated by the details associated with the technologies employed. (The third set of specifications is for photogrammetrically-derived floodplain mapping products.) During the related discussion on the topic, there was general consensus that it was not necessary to have separate technology-centric specifications for the floodplain mapping product. On the other hand, it would be valuable and necessary to have separate product-centric specifications that define the baseline performance criteria for acceptable digital elevation models to serve the separate applications of watershed mapping and floodplain mapping. Major Pitfalls to Avoid
There are two major pitfalls to be avoided in the development and use of standards and specifications for innovative remote sensing technologies. The first pitfall is in recognizing that there are many levels of mapping suitable for flood management. One level is the watershed, where IFSAR has been determined to be particularly effective. The second level is the floodplain, for which current sets of draft specifications have been developed, and where LIDAR has proven to be particularly effective. Additional specifications might be developed for control structures, dikes, or other flood barriers. The second, closely related pitfall in dealing with specifications such as FEMA 37, is to maintain focus on the fact that specifications should provide guidance for only very specific applications; in this case, the guidance is for NFIP studies and data production. As an application-centric specification, its requirements may not be appropriate to other applications. What is Accuracy
The current, draft version of the specification for flood mapping calls for 30cm vertical accuracy (approximately equivalent to 15cm percent RMSE) at the 95 percent confidence level. This performance requirement is consistent in each of the LIDAR, IFSAR, and photogrammetric specifications. Another topic being discussed within the community is related to the substantiation for this requirement for 30cm vertical accuracy.
In 1998, the Federal Geographic Data Committee (FGDC) published the National Standard for Spatial Data Accuracy (NSSDA) that supersedes the National Map Accuracy Standard (NMAS) of 1947, for digital products. The NSSDA replaces the NMAS for soft-copy (digital) spatial products (which includes DEMs and digital contours), but the NMAS remains in effect for hard-copy maps (which includes contours on published paper maps). The NSSDA states in FGDC-STD-007.1-1998, paragraph 1.2, Accuracy Standard, "The reporting standard in the vertical component is a linear uncertainty value, such that the true or theoretical location of the point falls within +/- of that linear uncertainty value 95 percent of the time." Note that the NSSDA defines accuracy at the 95 percent confidence level and assumes errors have a normal distribution, whereas the NMAS defines accuracy at the 90 percent confidence level and does not assume errors have a normal distribution. Thus, the 10 percent of points that fall outside the NMAS 90 percent standard could have errors of any magnitude, whereas the five percent of points that fall outside the NSSDA 95 percent standard should normally have errors only slightly larger than that standard, or otherwise those "outliers" would greatly magnify when squared as part of the RMSE computations. Whereas 90 percent and 95 percent appear to be similar, the former allows two "outliers" from 20 checkpoints, but the latter allows only one "outlier" from 20 checkpoints. Note also that the NMAS allows apparent vertical errors to be decreased by assuming a horizontal displacement within the permissible horizontal error for a map of that scale, whereas the NSSDA does not make such allowances. Many NMAS assumptions become meaningless with digital spatial data. Most DEM users have what they consider to be a clear concept of the meaning of vertical accuracy, or accuracy of ± one foot, for example, but such terms mean different things to different people. Few users realize that a Vertical Root Mean Square Error (RMSEz) of one foot nearly equates to Vertical Accuracy (Accuracyz) of two feet at the 95 percent confidence level. According to FGDC-STD-007.3-1998, Accuracy Statistics, NSSDA, Accuracyz = 1.9600 x RMSEz. It further states: "Accuracy of new or revised spatial data will be reported according to the NSSDA." A final pitfall to be avoided in evaluating LIDAR and IFSAR data is to ensure that everyone understands and uses the correct NSSDA terminology regarding accuracy of digital elevation data, including DEMs. Stating that something is "accurate to ±six inches" may not mean the same thing to others as it means to you. The accuracy needs to be stated in terms of either RMSEz or Accuracyz. Knowing the type of accuracy units and the relationship between different units will enable an understanding of their true meanings. References
"LIDAR and IFSAR: Pitfalls and Opportunities for Our Future," David Maune, Carlton Daniels, and James Dameron. ASPRS Conference Proceedings, May 2000. Back |
