A semi-automatic method to Hydro-flatten LiDAR data

Sagar S. Deshpandea, Alper Yilmazb aSurveying Engineering, Ferris State University, sagardeshpande@ferris.edu

bDepartment of Civil, Environmental, and Geodetic Engineering, The Ohio State University, Yilmaz.15@osu.edu

ABSTRACT

Triangulation of LiDAR data can produce unnatural water surface. Therefore, it is necessary to hydro-

flatten LiDAR data. Hydro-flattening is driven primarily by cartographic mapping needs, and can also

help hydrologic analysis, because hydro-flattened surfaces behave naturally; allowing water flow from

upstream to downstream. Detailed hydro-flattening requirements have been stated by USGS, which are

to be used by LiDAR contractors. In this paper, a new method is presented which extracts bank shoreline

of inland water bodies to hydro-flatten LiDAR data. Unclassified LiDAR data acquired over Big Rapids, MI

and a river centerline from the NHD were used as inputs to this method. LiDAR data within a certain

distance of the centerline was processed to create a continuous bare ground surface by removing LiDAR

points over trees and other non-ground features and by adding synthetic bathymetric points in the

water area with sparse LiDAR points. Then cross-sections were placed at regular intervals perpendicular

to the centerline, and the lowest elevation points around every intersection were searched. These

elevations were assumed to be at, or near the actual water surface and assigned to their respective

cross-section. These elevations were checked and revised to have a natural drop in elevation from

upstream to downstream, then raised arbitrarily by 0.5 m so that they were between the minimum and

the bank full elevation of the river at any cross-section. A virtual water surface was created using the

raised elevations, which was then intersected with the conditioned bare earth surface to obtain bank

shoreline polygons. Any voids greater than 200 m2 were filled and the polygons were smoothed using

tools in ArcGIS software for better cartographic appearance. These polygons were then converted to a

3D feature by extracting elevation from the virtual water surface and were used to hydro-flatten the

LiDAR data. The proposed method is also modified to be used for hydro-flatten ponds.

KEYWORDS: LiDAR, Hydro-flattening, bank shoreline, 3D breaklines

1. INTRODUCTION

Digital elevation model (DEM) or triangulated irregular network (TIN) derived from LiDAR data will show

unpleasant and unnatural water surface due to the fact that LiDAR data does not provide natural

breaklines and the elevations of water points is not uniform. Furthermore, hydrologic connectivity in

water surfaces is necessary to ensure that the elevation of water surface decreases from upstream (U/S)

to downstream (D/S). Therefore, the LiDAR data is hydro-flattened so that water bodies such as:

streams, rivers, and long reservoirs show an elevation change along their length, whereas ponds, lakes,

or other cartographically polygonal water surfaces to show a static water surface, consistent with their

natural behavior and the surrounding topography (Heidemann, 2012). Although hydro-flattening could

help in hydraulic and hydrologic modeling, the primary need of hydro-flattening is driven solely by

cartographic mapping needs. It is achieved by using a 3D breakline along the shoreline of a water body,

which establishes elevations of the water surface, and produce an aesthetically acceptable water

surface in the DEM or TIN.

The shoreline of a water body can be defined as the line of contact between land and water (Liu et al.

2009, Shalowitz, 1964) or “ . . . the intersection of the land with the water surface” (Gill and Schultz,

2001). The ordinary high water line is often used to demarcate property boundaries and has legal

significance. The ordinary high water mark or the average high water mark is defined as the high water

mark that can be expected to be produced by a body of water in non-flood conditions (USGS 2015a).

This high water line can be easily identified in aerial photographs as a wet-dry visible line in sandy

beaches (Liu et al. 2009, Shalowitz, 1964). In case of steep banks along river, the bank shoreline can be

represented by a line located between the water surface at the time of data acquisition and the high

water mark, and can be included in LiDAR data as breakline (Heidemann, 2012 and Maune, 2003).

Considering the total length of rivers, manual digitization of shorelines from aerial images would be

highly time-consuming. In addition, images alone cannot be used to determine shoreline in vegetated

areas. On the other hand, LiDAR data lacks feature information as it only records intensity and elevation

at discrete points and the presence of trees along the shore makes it difficult to identify the shoreline of

water bodies.

In this paper, a new method is presented to hydro-flatten LiDAR data. The main objective of this paper is

to develop a semi-automated method to minimize human interpretation and to increase time efficiency

and accuracy of hydro-flattening. The background information and the literature review are described in

Section 2. Section 3 describes the data used to test the proposed method. Section 4 describes the

methodology developed for hydro-flattening. The results are discussed in Section 5 followed by

conclusion in Section 6.

2. BACKGROUND REVIEW

With the improvements in the sensor technology over the past decade, there has been a significant

increase in the resolution and decrease in the cost of acquisition of LiDAR data (Chen, 2007 and

Deshpande, 2013). The basic principle of a LiDAR survey is based on laser distance measurement using a

scanning mirror mechanism. At a sampling rate of 30 kHz or higher, a LiDAR system can produce spatial

elevation data with an average horizontal spacing of 1 m or less (Liu et al., 2009); thus providing very

dense elevation data. LiDAR data can also measure ground elevation in vegetated areas, resulting in

increased details when compared to an aerial survey (Sasaki et al., 2008). This ability provides more

accurate topography along a river or stream in vegetated areas. Therefore, LiDAR is becoming

increasingly popular and is being used in different fields such as floodplain mapping, canopy mapping,

building extraction, etc.

In the USA, the 3D Elevation Program (3DEP) initiative is being developed by USGS in respond to the

growing needs for high-quality topographic data by collecting enhanced elevation data systematically in

the form of high-quality LiDAR data (USGS 2015b). It is anticipated that this program will provide

benefits to a range of Federal, State, and local government applications, and private industry

applications. Federal Emergency Management Agency (FEMA) is among the first to utilize LiDAR for

automated hydrologic and hydraulic (H&H) modeling and is the first federal agency to publish LiDAR

guidelines and specifications (Maune, 2003). Similarly, the USGS–NGP program provides leadership for

coordination, production, and service activities. USGS–NGP’s objective is to produce topographic DEMs

that resemble DEMs derived from traditional photogrammetric methods that are free of unnatural

triangulation effects over water surface (Heidemann, 2012). Some of the hydro-flattening requirements

which shall be achieved by the contractors are summarized below:

a) Inland ponds or lakes: Water bodies of 8,000 m2 (2 acres) or greater surface area at the time of

collection shall be flattened to have a single elevation for every shoreline vertex defining the

water body’s perimeter and that this elevation should be at or lower than the immediately

surrounding terrain.

b) Inland streams and rivers: Streams and rivers of a 30-m (100-ft) nominal width shall be flattened

to present a flat and level water surface bank-to-bank having a gradient downhill water surface,

following the immediately surrounding terrain. The entire water-surface edge shall be at or

below the immediately surrounding terrain.

In summary, the goal of hydro-flattening is to create an artificial water surface over water bodies which

will possess basic surface characteristics of natural water. Figure 1 shows the issues encountered while

identifying bank shorelines.

Figure 1: Schematic diagram showing interaction of LiDAR with ground features

Most airborne topographic LiDAR systems operate within the near-infrared spectrum, resulting in

reduced point density or void regions over water bodies (Figure 1) caused due to the absorption of laser

pulses by water (Worstell et al., 2014). The intensity value of these points is generally very low because

it is affected by factors such as velocity, turbulence, depth of the water, and poor reflectance

(Antonarakis et al., 2008). Both the intensity values and the point density of LiDAR over water bodies are

not consistent resulting in higher density and/or a wide range of intensity values. Further, several points

could be obstructed by overhanging trees or the presence of steep shore could also result in few or no

LiDAR points along banks (Liu et al. 2009, Maune, 2003), resulting in low point density. All the

aforementioned issues make identification of river or pond bank shoreline a challenging task.

Traditionally, two-dimensional breaklines derived photogrammetrically have been used to hydro-flatten

LiDAR data because it could be difficult to determine the shoreline location using LiDAR data due to

sparse points between the top and bottom of stream banks (Maune, 2003). Compared to

Photogrammetry, LiDAR laser pulse can pass through trees to measure the ground underneath, resulting

in more ground points beneath the trees. Although sparse, such points could help in locating the bank

shoreline.

Literature review

Worstell et al. (2014) have utilized the property of sparse point to identify voids in LiDAR data and

classified them as water. Results from this study indicated that a 5-meter radius moving window over a

1 m raster LiDAR grid with fewer than 23 returns (28 percent of the moving window) was sufficient for

delineating void regions. In their study, the average elevation minus 1 sigma was used to model the

elevation of lakes/ponds. Elongated features were classified as rivers and a least square approach was

implemented to assign elevation from downstream to upstream. Approach based on LiDAR point

density and average elevation was also used in Korzeniowska (2012) to locate water bodies. Tascano et

al. (2014) proposed a non-linear heuristic method to detect and delineate water bodies and generate

breaklines primarily in a rural area, using both LiDAR elevation and intensity data. The breaklines

generated using this method matched closely with the break lines drawn in a semi-automated way.

Smeeckaert et al. (2013) presented an automatic, efficient, and versatile workflow for land/water

classification of airborne topographic LiDAR points, using the Support Vector Machine (SVM) method to

classify points along the banks. Area along approximate bank shoreline was used to produce training

sample for the SVM classification. Poppenga et al. (2010) presented an approach to remove artificial

dams that are created by bridges, culverts, or other structures across rivers. Selective draining methods

that modify the elevation surface to drain a depression through an obstruction are presented. This

method is important in producing a complete hydro-flatten LiDAR surface. Höfle et al., (2009) presented

a new method for water surface mapping using the geometrical and intensity information of the LiDAR

points. Their method included two essential preprocessing steps; the correction of the signal intensity

and the modeling of laser shot dropouts. Vegetation which cover water bodies were removed using a

height threshold and the points were classified as water and land by codes 0 and 1, respectively. Then

the 0.5 contour, which would be between the coded points, was assumed as the breakline. A

plannimetric accuracy of 0.45 m was achieved by this method. In Antonarakis et al. (2008), the authors

classified open water by checking elevation information within 0.5 m from an approximate bare earth

and intensity value less than 55. In addition to LiDAR, authors have used SPOT satellite image to identify

water due to missing response over water body. Lin et al. (2008) proposed an algorithm with profile

factor as the kernel circle is proposed for automatic recognition of river using image unification, edge

extraction and skeleton generation. The LiDAR and remote sensing images were used to extract rivers.

Most of the above methods are based on the intensity and/or point density of LiDAR data, but both of

these properties are not consistent over water bodies. The intensity values lack calibration information

thus resulting in significant variation, whereas dense points can also be found at many locations over

water bodies. In addition, almost all the above methods convert the LiDAR point data to raster format.

Unlike a raster format, a TIN accurately maintains the exact features of the LiDAR data and is preferable

to a DEM when it is critical to preserve the precise location of narrow or small surface features such as

levees or narrow stream channels (Maune, 2001). The mapping accuracy of the TIN surface created by

using LIDAR data is much more accurate than alternative mapping accuracies on an interpolated 2 ft

contour surface, a 10 ft contour surface, or an NED 1 arc second DEM surface (Cohen, 2007).

In this paper, we present a new method which is based on the elevation of LiDAR point data. This

method was implemented directly on TIN surfaces, thereby avoiding errors introduced by conversion to

raster. The bank shoreline is expected to be located between the high and low bankline location as

shown in Figure 1. Use of such shoreline can be sufficient in hydro-flattening the LiDAR data

(Heidemann, 2012 and Maune 2003). We have tested the proposed method on the data discussed in the

next section.

3. DATA USED

The LiDAR data used in this study, was obtained from the GIS department of Mecosta County, MI. This

data was acquired by the Sanborn Mapping Co. on December 6, 2014 using Leica ALS70 system from a

flying height of 2500m. The vertical accuracy of 0.046 m RMSE was reported by the vendor. Attributes

such as ground class number, intensity, elevation, and return number were provided, but, only the

elevation attribute was used in this study. There were about 2 million points with a tile of 0.58 km2 area,

resulting in an average point density of 3.4 points/m2. In addition to the LiDAR, the NHD river centerline

was used as the approximate location of the river and orthoimages were used for result evaluation. The

orthoimages were acquired in 2010 at a spatial resolution of 0.3048 m and had a horizontal accuracy of

0.96 m. These images were also obtained from the GIS department of Mecosta County, MI.

4. METHODOLOGY

The entire workflow, as shown in Figure 2, can be divided into five steps: data preparation, virtual water

surface (VWS) creation, continuous bare ground surface (CBGS) creation, bank shoreline determination

and refinement, and hydro-flattening the LiDAR data using the refined bank shoreline. These steps are

described below in detail.

Step 1: Data preparation

The raw LiDAR data, the NHD river centerline, and the orthoimages were used as input data sources. The

centerline was compared manually to the orthoimages to verify that it was within the visible width (W)

of the river. To avoid processing over the entire extent, LiDAR data up to a distance of 1.5 ×𝑊 on either

side of the revised centerline was clipped and was used in the following processes. Here, W stands for

the average width of the river, estimated by visual inspection.

Step 2: Virtual water surface creation

A VWS is created commonly during hydraulic modeling of a river by connecting cross-sections which are

placed perpendicular to the river’s centerline. In this study, a similar VWS was created which had near-

water surface elevation that was gradually decreasing from upstream to downstream. In case of lakes,

the VWS had a constant elevation. This procedure is described below.

Figure 2: Steps to hydro-flatten LiDAR data

Cross-sections were placed perpendicular to the NHD river centerline at regular intervals (Figure 3a). To

obtain water surface elevation, it was assumed that the lowest elevation point around every cross-

section will be at, or near the actual water surface. Therefore, LiDAR data within a radius distance of R

from the intersection of river centerline and the cross-section was searched to find the lowest elevation.

Figure 3b shows the points within the search radius and the minimum point location. Due to varying

width of the river, 𝑅 = 1.5 ×𝑊 was adopted so that all the points between the banks can be searched.

Similarly, minimum elevation points around every intersection of the cross-sections and the river

centerline (Figure 3c) was identified and the minimum value was assigned to the respective cross-

sections.

a. Intersection of cross-

sections with centerline

b. Minimum LiDAR point within search

radius. Green points show the

location of the intersection of cross-

section and centerline

c. Minimum LiDAR point

searched around every

intersection

Figure 3: Process of searching minimum elevation LiDAR point

Figure 4 shows the elevation of the minimum elevation at every cross-section for a river reach of 7km.

Ideally, these elevations (shown as triangles) should have gradually increased from downstream to

upstream, but due to the random nature of the LiDAR point and the sparse point density over water, the

minimum elevation was not always the water surface elevation. Additionally, two low outliers were

observed at the intersections #18 and #63. To fulfill the requirement of hydro-flattening by maintaining

the hydrologic connectivity which ensures flow of water from upstream to downstream, elevations were

checked and modified prior to further processing.

Figure 4: Elevations assigned at the intersection of cross-sections and centerline

262

263

264

265

266

267

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81

Wat

er S

urf

ace

Elev

atio

n (

m)

Intersection of Cross-sections and centerline

Minimum Revised Intersection

The revised elevations are shown in the Figure 4 and were used at the respective cross-sections to

create a TIN surface. This surface is referred as Base-Virtual Water Surface (B-VWS) in the remainder of

the paper. As explained earlier, LiDAR points over the water surface are sparse and can produce some

vestige due to overhanging trees. From Figure 1, it can also be noted that there could be areas between

banks without any LiDAR points. These issues could create artificial islands between the banks if the B-

VWS is intersected with the bare ground surface because the B-VWS is too close to the actual water

surface. Thus, it was necessary to raise the B-VWS so that it is higher than the actual water surface but

lower than the bank-full elevation. Assuming that the elevation difference between the B-VWS and the

bank-full elevation at any location along the river is at least 0.5 m, a second VWS was created by adding

0.5 m to the B-VWS. This second surface is called as intersection-VWS (I-VWS) in the remainder of the

paper.

Step 3: Continuous bare ground surface (CBGS) creation

There are several methods of creating bare ground surface from LiDAR data (Sithole and Vosselman,

2004). In this paper, a simple approach proposed in Antonarakis et al. (2008) was adapted to create a

bare ground surface (Figure 2). A raster surface of 3 m grid size was created and the lowest elevation of

LiDAR points within each grid was used as the grid value by assuming that it belonged to the ground

surface. All LiDAR points within 0.5 m of this raster surface were classified as bare ground points,

whereas the rest were classified as non-ground points. We observed that several grids over the water

surface were without any point due to sparseness of LiDAR. These grids were labelled as “no elevation

grid.” Similarly, several other grids had very few points. Upon close inspection of such grids, it was found

that these points either belonged to the water surface, shallow underwater ground surface, or

overhanging trees. Therefore, grids with less than 3 points were identified and labelled as “no elevation

grid” and the points over these grids were classified as non-ground. At this stage, LiDAR data was

classified to have bare ground points, non-ground points, and no elevation grid. During further

inspection of the bare ground points, it was noticed that at several locations there were no points

between banks. A TIN surface created using the bare ground points could be entirely or partially above

the I-VWS, thus no intersection could take place between the two surfaces (Figure 5). It was necessary

to lower the elevation over no elevation grids so that the two surfaces could intersect.

Figure 5: Schematic diagram of a cross-section profile

Since the “no elevation grid” belonged to water area, points with significantly low elevation compared

to LiDAR data were placed at these grid locations to add synthetic bathymetry points. In this study, the

lowest elevation along the profile was 263m, so an elevation of 10 m was arbitrarily assigned to the

bathymetry points. This process was similar to river burning (Koltun, 2006) but was extended to a larger

area. Then a continuous bare ground surface (CBGS) was created by triangulating the bathymetry points

and the bare ground points.

Step 4: Bank shoreline extraction

In the bank shoreline extraction stage the CBGS was intersected with the I-VWS to obtain polygons of

the area below, equal, or above the I-VWS. The bank shoreline was identified using the polygons that

were below or equal to the I-VWS. Close inspection of these polygons revealed that few high points

were present within the banks resulting in small voids in the polygon. A smoothing and filtering process

was implemented to fill voids smaller than 200 m2 and to smooth the boundary using PAEK method with

a threshold value of 30.48 m (100 ft). This smoothed 2D bank shoreline was converted to 3D, using the

elevation from the B-VWS because this surface was near the actual water surface compared to the I-

VWS.

Step 5: Hydro-flattening

The last stage was to hydro-flatten the bare ground LiDAR data using the 3D bank shoreline polygon

obtained in the above process. The bare ground points within the smooth bank shoreline polygon were

classified as water points and those within 0.5m from the bank shoreline were classified as Ignored

ground (class value of 10) (Heidemann, 2012). Both the water points and the Ignored ground points

were not included in the triangulation process. All the remaining bare ground points were triangulated

using the 3D bank shoreline as a hardline.

Pond/Lake shoreline determination

The process explained above for a river was modified to hydro-flatten static water bodies such as ponds

or lakes. Instead of using a centerline, approximate pond boundary was buffered to process the points

within 10 m from the pond boundary. These points were processed to create the CBGS by following the

same procedure used for rivers. The B-VWS was created by using the minimum elevation within the

buffered points. In this case, the minimum elevation point was checked manually so that it is not a low

outlier. After obtaining the CBGS and VWS, the rest of the procedure remained the same as shown in

the Figure 2.

5. IMPLEMENTATION AND RESULTS

The above method was implemented on two different rivers: Muskegon River and Ryan creek having the

properties listed in Table 1. The shoreline derived by the proposed method and the vendor provided

shoreline are compared with the orthoimage in Figure 6. Figure 7a, c, and d show zoomed areas of

Muskegon River, whereas Figure 7b shows Ryan creek. These two shorelines were compared visually in

2D and also in 3D. It was observed that for most of the length both shorelines appear similar with

smaller variations. At several locations the proposed method shoreline appears to be landwards on

either bank compared to the vendor provided shoreline. This could be attributed to the fact that the I-

VWS was 0.5 m higher than the B-VWS, causing a landward intersection. At certain location in Figure 7a,

the proposed method shoreline was able to connected smaller adjacent water bodies which were

connected to the Muskegon River. The elevations along the shoreline was found to be gradually

decreasing from upstream to downstream Figure 7f.

Table 1: Properties of river reaches

Characteristics Muskegon river Ryan creek

Length processed 7.6 km (two reaches combined) 0.6 km

Visible width range 55 to 75 m 8-13 m (Not clearly visible)

Average width used (W) 61 m (200 ft) 11 m (50 feet)

Lowest elevation 262.9 m 263.7 m

Highest elevation 266.0 m 267.8 m

Average slope (V/H) 1/2100 1/150

The shorelines were also compared to GPS-RTK survey, conducted on February 22, 2016 at 41 locations along a 0.5 km reach of the Muskegon River using a Leica Viva GS14 receiver. The plannimetric location of the bank shoreline was measured during the survey. The horizontal precision of the survey was ±0.04 m. Statistics of the shortest distances from the GPS points to the two shorelines are shown in Table 2.

Table 2: Accuracy comparison of shorelines using GPS survey

GPS survey v/s vendor’s method GPS survey v/s proposed method

Minimum difference 0.03 m 0.02 m

Maximum difference 2.5 m 2.67 m

Mean difference 1.20 m 0.69 m

Standard deviation 0.60 m 0.57 m

The results showed that the mean deviation of the shoreline derived by the proposed method was 0.6m and the vendor provided shoreline was 1.20m. All other parameters were almost the same. It should be noted that the LiDAR was acquired in the month of December when the water level is typically lower than that in the month of February. The I-VWS, which was intersected with the CBGS, was created by adding 0.5 m to the B-VWS. Therefore, the lower mean difference of the proposed shoreline could be due to the water level in February being higher than that in December and closer to I-VWS.

Figure 6: Overview of the two rivers showing the shorelines

River reach of Ryan creek was also hydro-flattened using the same procedure. This creek is significantly

smaller and steeper compared to Muskegon River. The proposed method with the search radius and the

cross-section spacing of 1.5 ×𝑊 was able to extract bank shoreline effectively. After hydro-flattening,

the elevations along both rivers gradually drop from upstream to downstream, thus follow the

requirements stated by USGS.

The shoreline derived for the pond Figure 7e, also appears to be landwards compared to vendor

provided shoreline. The pond water surface elevation also fulfils the USGS requirements.

a. Zoom 1

b. Zoom 2

c. Zoom 3

d. Zoom 4

e. Shoreline of a pond

f. Hydro-flatten surface and

elevations drop from D/S to U/S Figure 7: Zoomed areas showing the comparison of the two shorelines

6. CONCLUSION

This paper has presented a new method to extract bank shoreline and perform hydro-flattening of

LiDAR data using the advantage of LiDAR that it can see through vegetation. The method is based on

intersection of CBGS and I-VWS to extract bank shorelines, which are then smoothened to use as

breaklines. The close inspection of the results in 2D and 3D visualization showed that this method is

capable of extracting bank shoreline closer to the actual bank shoreline location compared similar to

vendor provided shoreline. Further, the search radius and the cross-section spacing of 1.5 ×𝑊 was able

to extract bank shoreline effectively for both rivers despite the fact that they differ significantly in slope

and width. The B-VWS, which is used to extract height information, showed a gradual drop in elevation

conforming to the USGS requirement of Hydro-flattening. Similarly, the B-VWS of ponds showed even

elevation from bank to bank. Comparison with a field GPS survey showed that the proposed method

was able to achieve a higher accuracy of 0.69m compared to 1.20m for the vendor provided shoreline.

Finally, it was observed that despite all automation, a manual check (and correction if required) is

necessary because LiDAR points are not available at several highly vegetated locations.

Acknowledgements

We would like to thank the GIS department at Mecosta County, MI, State of Michigan – DTMB, and the

Sanborn Mapping Company for providing the LiDAR data and undergraduate students, Michelle Thebo,

Marshall Wixom, and Oliver Fom Kom for their help during the study.

References:

Antonarakis, A. S., Richards, K. S., & Brasington, J. (2008). Object-based land cover classification using

airborne LiDAR. Remote Sensing of Environment, 112(6), 2988-2998.

Cohen, C. (2007), The Impact of Surface Data Accuracy on Floodplain Mapping, CE 394K.3 GIS in Water

Resources, Final Report Fall 2007, University of Texas. Available at:

http://www.crwr.utexas.edu/gis/gishydro08/Introduction/TermProjects/Cohen.htm, accessed

March 2, 2011.

Deshpande, S. S. (2013). Improved Floodplain Delineation Method Using High‐Density LiDAR

Data. Computer‐Aided Civil and Infrastructure Engineering, 28(1), 68-79.

Höfle, B., Vetter, M., Pfeifer, N., Mandlburger, G., & Stötter, J. (2009). Water surface mapping from

airborne laser scanning using signal intensity and elevation data. Earth Surface Processes and

Landforms, 34(12), 1635-1649.

Heidemann, H. K. (2012). Lidar base specification version 1.0: US Geological Survey Techniques and

Methods, book 11, chap.

Koltun, G. F., Kula, S. P., & Puskas, B. M. (2006). A Streamflow Statistics (StreamStats) Web Application

for Ohio (No. FHWA/OH-2006/25). US Department of the Interior, US Geological Survey.

Korzeniowska, K. (2012). Modelling of water surface topography on the Digital Elevation Models using

LiDAR data. Jagiellonian University, Institute of Geography and Spatial Management, Cracow,

Poland.

Liu, J. K., Li, R., Deshpande, S., Niu, X., & Shih, T. Y. (2009). Estimation of blufflines using topographic

LiDAR data and orthoimages. Photogrammetric Engineering & Remote Sensing, 75(1), 69-79.

Maune, D. F. (2001), Digital Elevations Model Technologies and Applications: The DEM User Manual,

American Society of Photogrammetry and Remote Sensing, Bethesda, MD.

Maune, D. F. (2003). FEMA's Mapping and Surveying Guidelines and Specifications. In Proceedings of the

Fall 2003 Conference of the American Society for Photogrammetry & Remote Sensing.

http://www. dewberry. com/uploadedFiles/ASPRSFall2003FEMAStandards. pdf (05.10. 2004.).

Poppenga, S. K., Worstell, B. B., Stoker, J. M., & Greenlee, S. K. (2010).Using selective drainage methods

to extract continuous surface flow from 1-meter LiDAR-derived digital elevation data. U. S.

Geological Survey.

Smeeckaert, J., Mallet, C., David, N., Chehata, N., & Ferraz, A. (2013). Large-scale classification of water

areas using airborne topographic lidar data. Remote Sensing of Environment, 138, 134-148.

Sithole, G., & Vosselman, G. (2004). Experimental comparison of filter algorithms for bare-Earth

extraction from airborne laser scanning point clouds. ISPRS journal of photogrammetry and

remote sensing, 59(1), 85-101.

Toscano, G., Gopalam, U., & Devarajan, V. (2014). Auto Hydro Break Line Generation Using LiDAR

Elevation and Intensity Data. In Proceedings, ASPRS Conference, Louisville, Kentucky.

USGS (2015a). High-water marks and stream stage, URL:

http://water.usgs.gov/edu/highwatermarks.html, Page Contact Information: Howard Perlman,

Page Last Modified: Thursday, 30-Jul-2015 14:17:08 EDT

USGS (2015b). 3D Elevation Program (3DEP), URL: http://nationalmap.gov/3DEP/index.html, Page

Contact Information: 3D Elevation Program, Page Last Modified: Tuesday, 29-Sep-2015 10:41:34

EDT

Worstell, B. B., Poppenga, S., Evans, G. A., Prince, S., Lopera, J. G., Falendysz, E. A., and Williamson, J. E.

(2014). Lidar point density analysis: implications for identifying water bodies (No. 2014-5191).

US Geological Survey.