The State of North Carolina’s Floodplain Mapping Program is the result of the State’s designation by the Federal Emergency Management Agency (FEMA) as the first Cooperating Technical State (CTS).  Under this designation, North Carolina has full responsibility over the Flood Insurance Rate Maps (FIRMs) within its state boundaries. In carrying out this responsibility, the State of North Carolina is showing leadership in producing specifications for advanced surveying techniques needed for LIDAR derived products such as the digital terrain model (DTM). 

An important application for LIDAR derived DTMs is the prediction of flood risk using hydraulic and hydrologic (H&H) models.  The State of North Carolina is in the process of using LIDAR to create better flood plain models than are currently available from other sources.  Development of this technology is of increasing importance to other local, regional, and state governments.  H&H (especially hydraulic) models depend on high DTM accuracy.  In the case of flood risk mapping, large dollar amounts associated with development planning, insurance, and flood disaster recovery make the DTM accuracy issue paramount.  What LIDAR acquisition parameters are required for a given DTM accuracy?  What zoning errors will result from DTMs of various accuracies?

Currently there are few guidelines to answer these questions.  Few studies (North Carolina, 2002; Hodgson, et al., 2003) have attempted to document the accuracy of digital terrain models (DTMs) derived from LIDAR.  Fewer studies have examined the effect of DTM accuracy on hydrologic modeling (Kenward, et al., 2002).  We are not aware of any studies that have researched the effects of DTM accuracy on flood zoning errors. 

Previous research has established that DTM accuracy is affected by land cover type (North Carolina, 2002; Hodgson, et al., 2003).  Without above-ground obstructions more LIDAR returns are from the ground and therefore the spatial density of observations is greater.  Also, the “vegetation removal” process for labeling LIDAR returns is less accurate in forested land cover.  To mitigate against the over-story obstruction problem, LIDAR data at higher posting densities can be collected. 

Posting density is especially important because it is the greatest cost factor in LIDAR data acquisition / processing and because it often improves accuracy.  A higher posting density requires a more sophisticated sensor system with a higher pulse rate (e.g. 50,000 pulses per second), lower elevation flights (and therefore more flight lines), a narrower scan angle, or a combination of these.  These cause acquisition for a given aerial coverage to become more expensive.  Beyond acquisition costs, significantly more computing resources (processor speed, RAM, storage space, etc.) and technical personnel time are required to process higher posting densities during the DTM creation process.  Can a balance between LIDAR accuracy and cost be revealed by identifying the optimal posting density for a given physiography?

H&H models depend on suitable accuracy, and of course it is preferable if the costs of developing these models are minimized.  In hydrologic modeling applications, there is evidence that higher resolution DTMs (i.e., from high posting densities) result in estimates of higher mean slopes.  In other words, the modeled mean slope of a basin will be greater for a DTM created from 1 × 1 m LIDAR data than from 5 × 5 m data.  Is this functional relationship predictable?  In hydraulic applications, higher resolution DTM data means that smaller physiographic features are identifiable.
 

References

Hodgson, M.E., J.R. Jensen, L. Schmidt, S. Schill and B. Davis, 2003, "An Evaluation of LIDAR- and IFSAR-derived Digital Elevation Models in Leaf-on Conditions with USGS Level 1 and Level 2 DEMs", Remote Sensing of Environment 84: 295-308.

Kenward, T., D.P. Lettenmaier, E.F. Wood and E. Fielding, 2000, "Effects of Digital Elevation Model Accuracy on Hydrologic Predictions", Remote Sensing of Environment 74: 432-444.

North Carolina Floodplain Mapping Program, 2002, http://www.ncfloodmaps.com/pubdocs/NCFPMPHndOut.htm.