A 1.410Mb PDF of this article as it appeared in the magazine complete with images is available by clicking HERE
In 1998, Dewberry was one of four prime contractors who won USGS’ Cartographic Services Contract (CSC) for which USGS had made no reference to LiDAR in its Request for Proposal. We were poised to produce digital orthophotos, Digital Line Graphs (DLGs), Basic Graphic Revisions (BGRs), National Hydrography Datasets (NHDs) and conventional DEMs (from cartographic sources), but there was no request for LiDAR services.
In August of 1998 I received a phone call from Phil Havens, USGS’ COR for the CSC contract, asking: "Dave, do you know anything about LiDAR?" I replied that Dewberry had worked on LiDAR projects for our contracts with FEMA and the Mobile District of USACE and I had advised NOAA on the benefits of LIDAR when I authored the National Height Modernization Study on how to modernize the National Height System in the U.S. I explained some of the capabilities and limitations of different LiDAR sensors. Phil liked that I was familiar with LiDAR, and we took the necessary steps to provide LiDAR services to USGS.
USGS subsequently issued Dewberry its first CSC task order for LiDAR data of Snoqualmie, WA, and Morrison, CO–more to see what LiDAR was capable of doing than to meet a strict specification. Snoqualmie was chosen because photogrammetric contours had never previously been produced there because of the tall, dense forests; and Morrison was chosen because it had been accurately surveyed and mapped on prior occasions and could be used as a calibration site. The Morrison project proceeded very well, but I recall two major issues with Snoqualmie: (1) persistent cloud cover for 10-11 months, and (2) difficulty penetrating the dense vegetation. Back then, USGS was primarily interested in DTMs and didn’t worry about DSMs. DTM artifacts were the primary issue for years when vegetation and manmade features were "removed" manually from the bare-earth DTM.
During the next two years, we executed nine LiDAR task orders under the CSC contract, winning a special award from USGS "for outstanding achievements in producing LiDAR products of the highest quality in a timely manner." Now, in our 5th consecutive USGS contract, we have executed 68 LiDAR task orders to date. I credit FEMA for giving me the opportunity to learn the capabilities and limitations of LiDAR in the years immediately preceding this USGS contract–and already knowing how to implement a LiDAR task order and QC the data because, at that time, USGS had no LiDAR specifications and wasn’t sure what to ask for. For over a decade, USGS relied partly on FEMA’s LiDAR guidelines and specifications or ad hoc Scopes of Work. In 2004, as FEMA’s representative on the Technical Working Group of the National Digital Elevation Program (NDEP), I was the primary author of the NDEP Guidelines for Digital Elevation Data used by USGS until 2015 to test and report LiDAR’s Fundamental Vertical Accuracy (FVA) in open terrain, Consolidated Vertical Accuracy (CVA) for all land cover categories combined, and Supplemental Vertical Accuracy (SVA) for individual land cover categories.
In 2010, USGS published its draft LiDAR specification which became official in 2012 with the USGS Lidar Base Specification V1.0. Today, we use the USGS LiDAR Base Specification V1.2 (Figure 1) for nationwide Quality Levels QL1 and QL2 LiDAR, with newer specifications planned for the future to include Geiger mode and single photon LiDAR. Both of these Quality Levels (QLs) require RMSEz <10 cm. For point density, QL1 has 8 points/m2 and QL2 has 2 points/m2. QL2 or better is the national standard for the 3D Elevation Program (3DEP).
Since those three years for Dewberry’s initial CSC contract with USGS, we have continued to execute LiDAR task orders for the CSC2 contract and all three of USGS’ Geospatial Products and Services Contracts (GPSC), using different (mostly small business) subcontractors for acquiring LiDAR data, with Dewberry normally processing the data, conducting checkpoint surveys and QA/ QCing the data in-house for consistency.
As part of our GPSC2 contract, I was privileged to be asked to manage the project and author the National Enhanced Elevation Assessment (NEEA) report which provided the blueprint for USGS’ 3DEP based on QL2 LiDAR or better for 49 states and QL5 IFSAR for Alaska. Working with USGS, I coined the terms and definitions for QL1, QL2, and QL3 LiDAR, and QL4 and QL5 DEMs used throughout the NEEA. Today, I am extremely pleased that USGS is strongly focused on acquiring high quality QL1 or QL2 LiDAR nationwide for the 3DEP, doing so in a timely and cost-effective manner–and I’m pleased that other agencies and states are joining in pursuit of this common goal. Future articles in LiDAR Magazine will review the NEEA and 3DEP in greater detail.
I will summarize several of our USGS LiDAR task orders that I consider to be memorable because of special challenges.
Among our memorable task orders for LiDAR services:
Our NRCS DEM Whitepaper (2011) identified NRCS user requirements and benefits for LiDAR data and served as a pilot for a broader study to come with the NEEA.
Our NEEA study identified LiDAR requirements and benefits nationwide for 34 federal agencies, all states and territories and numerous non-governmental organizations. Analyzing 602 mission critical activities, the NEEA was the most comprehensive benefit/cost analysis ever performed for any layer of The National Map, and it documented a minimum return on investment of 5:1, i.e., $5 in benefits for every $1 spent on LiDAR.
The NEEA subsequently led to USGS’ National Hydrography Requirements and Benefits Study (NHRBS), also executed by Dewberry under the GPSC2 contract which, among other benefits, documented the linkage between hydrography and elevation data
Our evaluation of Geiger mode and single photon LiDAR allowed USGS to determine their suitability for 3DEP requirements.
Our consulting on topobathymetric LiDAR has enabled USGS to acquire and/or process topobathymetric LiDAR of coastal areas.
Among our memorable task orders for LiDAR products:
The National Park Service actually stopped the flow of the Colorado River for several days so we could map the Grand Canyon’s topography mostly free of water.
LiDAR task orders for NGA required, extensive breaklines, generation of 2-D and 3-D building footprint shapefiles, forest polygons and individual tree points.
LiDAR data acquisition along the entire Texas/Mexico border was politically complicated.
Four LiDAR task orders involved wildfires and low-altitude LiDAR acquisition under dangerous conditions.
Tide-coordinated LiDAR acquisition of the San Francisco Bay was challenging as was the Channel Islands National Park with severe access rights by land and air, to avoid disturbances to nesting birds (Figure 2).
LiDAR of the Salton Sea area was complicated because of geological factors and unreliable control in the area.
Our high-density QL1 LiDAR of Louisa, VA helped geologists identify seismic fault lines following the 2011 earthquake that impacted central to northern Virginia and Washington, D.C.
Our helicopter-based LiDAR of the entire Pacific coastline will be vital for a special El Nio study.
Major differences between then (1998) and now (2016):
In 1998, we had no established methodology for assessing absolute vertical accuracy of LiDAR. Today, USGS follows the ASPRS Positional Accuracy Standards for Digital Geospatial Data that computes Non-vegetated Vertical Accuracy (NVA) using RMSEz, and Vegetated Vertical Accuracy (VVA) using the 95th percentile. Both QL1 and QL2 LiDAR are based on RMSEz of 10 cm or less in open terrain, whereas in 1998, we didn’t know if 20 cm RMSEz was achievable.
In 1998, we had no methodology for assessing the relative accuracy of LiDAR swaths. Today, USGS has specifications for within-swath hard surface repeatability and swath-to-swath RMSD and maximum differences for points in non-vegetated terrain mapped with overlapping swaths.
For years, LiDAR data quality was defined in terms of cleanliness from artifacts; we had no objective way to determine if an area was 95% clean from artifacts, for example (Figure 3). Today, we expect that LiDAR datasets will be free of errors that would cause an impact to the usability of the LiDAR data. For example spikes, divots, buildings, vegetation, and other significant above ground features should be corrected regardless of the percentage of points they impact. Minor issues such as one or two points being classified as class 1 (unclassified) vs class 2 (ground) in an area with sufficient point density would be acceptable and considered to be part of the allowable 1% of error within a 1 sq km area.
In 1998, LiDAR nominal pulse spacing was 5 to 10 meters between points. Today, QL1 and QL2 LiDAR have sub-meter point spacing with 8 and 2 points/m2 respectively, and even much higher point densities collected from Geiger mode and single photon LiDAR with 50 points/m2 or more.
Until 2007, I thought it was impossible for a second laser pulse to be emitted and tracked until the first pulse returned, but then multi-pulse in the air (MPIA) was introduced, allowing us to fly higher.
In 1998, automated feature extraction from LiDAR was mostly a dream; today it is commonplace. In 1998, LiDAR sensors acquired several thousand pulses per second; today they acquire several hundred thousand pulses per second.
The ASPRS LAS laser file format V1.0 was not developed until 2003. Now we are using LAS specification V1.4, and we classify and retain all laser returns in the point cloud. Prior to the LAS specifications, we "removed" points on vegetation and manmade features and wasted many of the returns that provided valuable information about the vegetation canopy and understory and the top surfaces of manmade features.
In 1998, about 10% of the postprocessing was automated; today about 90% is automated.
In 1998, we acquired low-density LiDAR of relatively small areas and paid several thousand dollars per square mile for it. Today, we acquire high-density LiDAR over large areas and pay several hundred dollars per square mile for it.
Prior to 2012, with no common specification, most agencies acquiring LiDAR data defined their own specifications, and "menus" of options were used to tailor specifications for diverse applications. Today, the USGS Lidar Base Specification has become the national standard, and it is advantageous to everyone to conform to this specification for either QL1 or QL2 LiDAR.
For many years, LiDAR was assumed by USGS to refer to topographic LiDAR; today we acquire and/or process topobathymetric LiDAR of coastal areas and merge it with topographic LiDAR.
Today, all of USGS’ GPSC3 prime contractors are focused on delivery of QL1 or QL2 LiDAR for the 3DEP which was fathered by the NEEA. Both the NEEA and 3DEP will be featured in subsequent columns by the author.
Dr. David Maune is an Associate Vice President at Dewberry where he is an elevation specialist and manages photogrammetric, LiDAR, IFSAR and sonar projects for USGS, NOAA, FEMA, USACE, and other federal, state and county governments. He has served as Dewberry’s Senior Project Manager for all USGS contracts since 1998. He specializes in independent QA/QC of LiDAR data produced by others and is perhaps best known as the editor and primary author of the 1st and 2nd editions of Digital Elevation Model Technologies and Applications: The DEM Users Manual published by ASPRS. He is a retired Army Colonel, last serving as Commander and Director of the U.S. Army Topographic Engineering Center (TEC), now the Army Geospatial Center (AGC).
A 1.410Mb PDF of this article as it appeared in the magazine complete with images is available by clicking HERE