Lidar Analysis in ArcGIS 10 for Forestry Applications

An Esri

White Paper • January 2011

Lidar Analysis in ArcGIS

for Forestry Applications

Esri, 380 New York St., Redlands, CA 92373-8100 USA

TEL 909-793-2853 • FAX 909-793-5953 • E-MAIL [email protected] • WEB esri.com

Printed in the United States of America.

The information contained in this document is the exclusive property of Esri. This work is protected under United States

transmitted in any form or by any means, electronic or mechanical, including photocopying and recording, or by any

information storage or retrieval system, except as expressly permitted in writing by Esri. All requests should be sent to

Attention: Contracts and Legal Services Manager, Esri, 380 New York Street, Redlands, CA 92373-8100 USA.

The information contained in this document is subject to change without notice.

Esri, the Esri globe logo, ArcGIS, 3D Analyst, ArcMap, ArcEditor, ArcToolbox, ArcCatalog, ArcInfo, ArcGlobe, esri.com

and @esri.com are trademarks, registered trademarks, or service marks of Esri in the United States, the European

Community, or certain other jurisdictions. Other companies and products mentioned herein may be trademarks or registered

trademarks of their respective trademark owners.

J-9999

Esri White Paper i

Lidar Analysis in ArcGIS 10

for Forestry Applications

An Esri White Paper

Contents Page

Executive Summary.............................................................................. 1

Keywords .............................................................................................. 1

Author ................................................................................................... 1

Introduction........................................................................................... 2

What Is Lidar?....................................................................................... 2

Advantages to the Forest Industry ........................................................ 3

Managing and Understanding Lidar Data............................................. 5

Understanding Raw Lidar Data ...................................................... 5

Point File Information Tool ...................................................... 5

Lidar Classification in ArcGIS ................................................. 7

Loading the Lidar Files to ArcGIS ................................................. 8

LAS to Multipoint Tool ............................................................ 9

Visualizing and Storing Lidar Data with ArcGIS................................. 13

Visualizing Lidar Data.................................................................... 13

Advantages of a Raster ............................................................. 14

Advantages of a Geodatabase Terrain ...................................... 14

Building and Delivering DEMs and DSMs from Lidar........................ 15

The Workflow to Create a Terrain and Deliver to Clients.............. 15

Building a Geodatabase Terrain...................................................... 16

Building a Raster DEM................................................................... 21

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

January 2011 ii

Contents Page

Analyzing Lidar Data for Foresters ...................................................... 24

Calculating Vegetation Characteristics from Lidar Data................ 24

Tree Height Estimation............................................................. 24

Biomass Density Calculation.................................................... 26

Point to Raster Tool ............................................................ 26

Replacing NoData Values as Zero Vegetation

Density .............................................................................. 26

Merging the Aboveground and Ground Results ................. 28

Creating a Floating Point Raster File.................................. 28

Calculating Density............................................................. 28

Distributing Large Lidar Datasets......................................................... 30

Preparing Raster DEM for Serving with the ArcGIS Server

Image Extension............................................................................ 30

Serving an Elevation Service through the ArcGIS Server

Image Extension............................................................................ 31

Creating an Elevation Mosaic Dataset............................................ 33

Visualizing an Elevation Service.................................................... 37

Creating a Referenced Mosaic Dataset ........................................... 38

Applying a Mosaic Function........................................................... 40

Estimating Tree Height Using Mosaic Dataset Functions.............. 43

Creating the Height Estimation Mosaic Dataset............................. 43

Applying the Arithmetic Function.................................................. 44

Conclusion ............................................................................................ 48

Acknowledgments................................................................................. 48

J-9999

Esri White Paper

Lidar Analysis in ArcGIS 10

for Forestry Applications

Executive Summary

Foresters use light detection and ranging (lidar) data to understand the

forest canopy and terrain, which helps them with forest management and

operational activities. Combining lidar data with Esri

ArcGIS

helps

analysts assess forest health, calculate forest biomass, classify terrain,

identify drainage patterns, and plan forest management activities such as

fertilization, harvesting programs, development activities, and more.

This paper will step through processes to convert lidar data into a format ArcGIS can

process, explain methods to interpret the lidar data, and show how ArcGIS can

disseminate the data to those who are not geospatial analysts. It will present methods for

reading raw classified lidar data and demonstrate methods for

■ Analyzing and validating lidar data with ArcGIS before any extensive processing

occurs

■ Storing and managing millions or billions of lidar points within the geodatabase in a

seamless dataset, regardless of the number of original lidar files

■ Processing to extract digital elevation models (DEMs) and digital surface models

(DSMs) from the lidar data and store them as terrains in a geodatabase or as raster

elevation files

■ Extracting vegetation density estimates and tree height estimates from lidar, which

aid in growth analysis, fertilization regimes, and logging operations

■ Serving and analyzing large amounts of lidar data as a seamless dataset to

geographic information system (GIS) clients

In all areas, ArcGIS is a complete system for managing, storing, and analyzing lidar data.

Coupling ArcGIS Desktop with ArcGIS Server, the forestry professional is able to access

large amounts of lidar data quickly and efficiently without the need to produce additional

resultant datasets.

Keywords

Lidar, ArcGIS, terrains, geodatabase, mosaic dataset, ArcGIS Server Image extension

Author

Gordon Sumerling, ESRI Australia Pty. Ltd., Adelaide, South Australia

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

January 2011 2

Introduction

ArcGIS can be used to analyze and manipulate lidar data to provide useful results for the

end user. This paper provides the processes to analyze and manipulate lidar data and

details how to

■ Check the supplied data.

■ Read and separate the point cloud data into ground and canopy returns.

■ Pass the resultant point clouds to a terrain dataset that creates a viewable and

displayable surface.

■ Perform analysis on the terrain dataset for tree height delineation and canopy density.

■ Pass the terrain data to the ArcGIS Server Image extension for dissemination to a

wider audience as a seamless viewable surface that can be accessed from GIS

technology.

What Is Lidar?

Lidar stands for light detection and ranging. In its most common form, it is an airborne

optical remote-sensing technology that measures scattered light to find range and other

information on a distant target. Similar to radar technology, which uses radio waves, the

range to an object is determined by measuring the time delay between transmission of a

pulse and detection of a reflected signal. Instead of radio waves, lidar uses much shorter

wavelengths of the electromagnetic spectrum, typically in the ultraviolet, visible, or near-

infrared range.

This technology allows the direct measurement of three-dimensional structures and the

underlying terrain. Depending on the methodology used to capture the data, the resultant

data can be very dense, for example, five points per meter. Such high resolution gives

higher accuracy for the measurement of the height of features on the ground and above

the ground. The ability to capture the height at such high resolution is lidar's principal

advantage over conventional optical instruments, such as digital cameras, for elevation

model creation.

In addition to the height attribute being captured, each return can capture other attributes

such as the return intensity, return number, and number of returns. The existence of these

attributes is dependent on the lidar data supplier and the supply order. These additional

attributes can all be used in the analysis of lidar data.

The intensity value is a measure of the return signal strength. It measures the peak

amplitude of return pulses as they are reflected back from the target to the detector of the

lidar system. Intensity is often used as an aid in feature detection and, where conventional

aerial photography is not available, can be used as a pseudo-image to provide the context

of the lidar acquisition area. See the image below:

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

Figure 1

Lidar Intensity Image

Lighter areas represent strong returns. Darker areas represent weaker or partial returns.

In forestry, lidar can be used to measure the three-dimensional structure of a forest stand

and produce a model of the underlying terrain. The structure of the forest will typically

generate a first return from the uppermost limit of the canopy, followed by less intense

returns through the canopy, down to the underlying terrain. Returns are classified into

ground and aboveground sources. The ground returns can generate a detailed terrain of

the area of interest, while the canopy returns can be filtered to provide forest structure at

the canopy and middle level of the forest.

Advantages to the

Forest Industry

The ability to simultaneously visualize the ground and model the canopy structure

provides significant advantages to the forest industry. Traditionally, foresters and land

managers have relied on topographic maps for terrain classification and field-based

surveys to obtain tree volumes and height information. Lidar data provides significant

improvements over both these techniques.

Existing topographic maps depict contours and rivers, which have been, for the most part,

captured from aerial photography using stereographic terrain generation techniques. In

areas where the tree canopy obscures the underlying terrain, interpretive methods are

used to depict where streams and contours occur. Terrains generated from lidar data more

accurately represent these geographic features. Lidar penetrates the tree canopy to return

a more accurate interpretation of the ground surface. This increases the accuracy of

terrain classification and thereby the resultant interpretation and analysis of the

geographic features.

Esri White Paper 3

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

January 2011 4

Lidar has provided significant benefits for forest development and engineering operations

including locating roads, harvest planning, forest regeneration, and more. The ability to

identify suitable creek crossings, determine optimal routes, and locate previously

unmapped historic roads aids in reducing costs and creating operational efficiencies.

Lidar has also offered an improvement to existing forest inventory methods and

procedures. Traditional field-based timber inventory methods are based on measurements

derived from systematically sampling plots in forest stands. This statistical sampling

method is most often used in forests where measuring every tree is impractical. Tree

volumes and heights are calculated in each sample plot, then generalized throughout a

forest stand that shares similar characteristics. Estimated results help describe stand

characteristics but are inaccurate due to variability in growing conditions throughout the

forest, sampling bias, and lack of precision. In addition, the time to collect such

measurements is both lengthy and expensive, as many sample plots may be required to

provide a reliable representation. Lidar can overcome these limitations.

An increasing number of forestry and land management organizations are using lidar for

forest inventory measurements. A wide range of information can be directly obtained

from lidar including

■ Digital elevation models

■ Tree heights and digital surface models

■ Crown cover

■ Forest structure

■ Crown canopy profile

Postprocessing of lidar data can reveal

■ Volume—Canopy geometric volume

■ Biomass—Canopy cover

■ Density—Height-scaled crown openness index and counts of delineated crowns

■ Foliage projected cover—Crown dimensions

The forest industry is requiring increasingly precise inventories to guide forest

management activities. Using lidar data, forest inventories can be conducted at nearly the

single tree level, offering more accurate representations of the true forest stand structure.

For forest inventory activities, lidar has been used primarily to retrieve basic structural

tree attributes including height, canopy cover, and vertical profiles. These attributes can

be used to derive other critical forestry measurements including basal area and timber

volume, as well as biomass for alternative energy and carbon sequestration analysis. This

paper will address these attributes.

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

Managing and

Understanding

Lidar Data

Understanding Raw

Lidar Data

Before any analysis is performed with lidar data, the data received must be checked for

any inconsistencies. Lidar data can be delivered in either binary .las format or ASCII

.xyz files. The LAS file format is a public binary file format, developed by the American

Society for Photogrammetry and Remote Sensing (ASPRS), that is an alternative to

proprietary systems or a generic ASCII file interchange system used by many data

providers. Details on the format can be found at asprs.org/society/committees/lidar/

Although a data provider will endeavor to provide the best quality data to its clients, there

is always a chance a client will encounter anomalies in the data. These can be in the form

of irregular minimum bounding shapes or holes in the sampling. It is therefore necessary

to check the quality of the data before performing any analysis.

The Point File Information tool in Esri's ArcGIS Desktop 3D Analyst

™

assists in

performing data quality assurance checks.

Point File

Information Tool

The Point File Information tool, found in the 3D Analyst toolbox in ArcGIS

(ArcToolbox\3D Analyst Tools\Conversion\From File\Point File Information), reports

important statistics about the raw lidar files.

The tool is designed to read the headers of LAS or scan ASCII files and summarize the

file contents. As a single lidar file often contains millions of points and many lidar

datasets contain more than one file, the Point File Information tool can accommodate

reading one or more files by specifying either individual lidar files or folders.

Esri White Paper 5

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

The result from this tool is a feature class that shows the

■ Minimum bounding rectangle for each file

■ Number of points recorded

■ Average point spacing

■ Minimum/Maximum z-values

When the feature class is loaded into ArcMap

™

, the minimum bounding rectangle of each

lidar file is drawn. Lidar data files are usually uniform in size, so if any of the feature

shapes appear large or irregular compared to the majority of features from the feature

class, they will appear different in ArcMap, which will alert an analyst, who can then

refer to the corresponding lidar data file for further investigation.

The average point spacing is important and should be uniform throughout the data files.

If any of the files have an average point spacing that is significantly larger than other

files, this may indicate incorrect sampling. In addition, average point spacing is important

when building geodatabase terrains and converting lidar files to feature classes.

The average point spacing is a product of the total number of points divided by the area

of the lidar data file. In cases where a lidar data file is only partially covered by points,

such as along a coastline, the average point spacing will be calculated to be greater than

the point spacing of the area sampled. These anomalous files would still be used in the

dataset, but their calculations would be excluded from further processing.

January 2011 6

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

The images that follow show three of the elements as reported by the Point File

Information tool, including

■ A uniform grid showing the extents of each lidar file.

■ The attribute table associated with the lidar extents, showing the average point

spacing, point count, minimum and maximum z-values, and originating file names.

■ The average point spacing as indicated by the statistics from the Pt_Spacing column.

In this example, the average point spacing tends to be approximately 0.6 meter. The

lidar dataset used in this paper was captured at a sampling density of two returns per

square meter; thus, 0.6 meter gives a good approximation to the ordered capture rate.

Again, if there were any significant outliers in the files, these would be highlighted

for further inspection.

Lidar Classification in

ArcGIS

LAS files contain a classification field that identifies each point's return type. This is

known as the class code. A classification describes the point return as Ground returns,

Canopy returns, Building returns, or Unclassified. This is useful in determining the

content of each lidar file.

Esri White Paper 7

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

Having the classification field available as part of the tool immediately identifies whether

the lidar file has been classified and whether it can be used for interpreting the terrain or

forest structure. If no classifications exist, either there is a problem with the file, which

may need to be updated, or the lidar file has not been classified at all. The field also helps

the analyst when interpreting data where no documentation exists. It provides a good

understanding of the file's content and how it has been classified.

ArcGIS reads the classification field with the Point File Information tool. Toggling the

Summarize by class code causes the tool to scan through the LAS files and analyze the

class code values. The attribute table of the output feature class will contain statistical

information for each class code encountered.

Loading the Lidar

Files to ArcGIS

Lidar data is characterized by very dense collections of points over an area, known as

point clouds. One laser pulse can be returned many times to the airborne sensor. A pulse

can be reflected off a tree's trunk, branches, and foliage as well as reflected off the

ground. The diagram below provides a visual example of this process.

January 2011 8

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

These multiple returns create a data management challenge. A single lidar file can

typically be 60 MB to 100 MB in size and can contain several million points. If this data

is loaded directly into a table, it creates many millions of records, which results in a large

data file that becomes difficult to manage. This challenge is overcome by loading the

points into the geodatabase feature type known as multipoint.

Multipoints are used to store thousands of points per database row. This is achieved by

storing the geometries in a Shape field and optional attributes in an Esri binary large

object (BLOB) field. The Shape field and the BLOB field are collections of binary data

stored as a single entity in a database. Storing the data as multipoints allows optimized

compression of shapes, reduces storage requirements, and improves database

performance. Any tool written to exploit these fields needs to understand the Esri BLOB

and Shape fields structures.

The tool to load lidar files into the geodatabase is called LAS to Multipoint. It is part of

the 3D Analyst toolset in ArcGIS (ArcToolbox\3D Analyst Tools\Conversion\From

File\LAS to Multipoint).

LAS to Multipoint

Tool

The LAS to Multipoint tool enables the user to read the lidar data files and load them into

the geodatabase. Many lidar analysis applications on the market today can perform

detailed analyses against lidar files, but only on individual files. Loading the lidar files in

a geodatabase allows a seamless mosaic of the entire lidar dataset, which then can be

analyzed by ArcGIS tools.

Esri White Paper 9

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

January 2011 10

When using the LAS to Multipoint tool, all the points can be loaded into the geodatabase.

This is useful for producing a point density map; however, this is not useful for canopy

and ground analysis. For this type of analysis, it is better to separate data into unique

classifications.

LAS files captured since September 2008 should conform to the LAS 1.2 specification.

This specification allows the separation of lidar data into ground returns and nonground

returns by the classification field. A full description of the specification can be found at

asprs.org/society/committees/standards/LiDAR_exchange_format.html

Lidar datasets captured prior to this will often contain the classification in the metadata

that is associated with lidar data.

The table below is an extract from the specification and describes the classification codes.

When lidar data is provided as part of a data order, the classifications would normally be

provided as part of the delivered documentation.

Classification Value Description

0 Created (never classified)

1 Unclassified

2 Ground

3 Low Vegetation

4 Medium Vegetation

5 High Vegetation

6 Building

7 Low Point (noise)

8 Model Key Point (mass point)

9 Water

When the LAS data files are read by the LAS to Multipoint tool, it can accommodate

these classifications and separate them into unique feature classes.

The LAS 1.2 specification also defines how to separate returns from the first return

through the last return. The return value is stored in the LAS file with the point

information. Having these values enables the extraction of the upper canopy data based

on the first reflected values. Midcanopy and ground values are reflected at any time.

The LAS to Multipoint tool needs certain specifications depending on the project. In the

following screen capture

■ A folder is specified for the data source. (Individual files can be specified, but when

large amounts of lidar data are being read, then it is a best practice to specify the

folder.)

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

Esri White Paper 11

■ The Output Feature Class is specified in a file geodatabase. (Multipoint feature

classes exist in geodatabases and shapefiles, but geodatabases are preferred due to

the extended capabilities of the geodatabase and the size restrictions of a shapefile.)

■ The ground spacing is specified. (This was acquired from the Point File Information

tool or from the supplied metadata.)

■ The Input Class Code to be extracted is specified. (The specific code entered will

vary depending on the classification being analyzed.)

■ The returns are focused on the ground returns.

■ The LAS file extension is designated as a .las file.

■ The coordinate system is specified.

Other specifications may also be considered:

■ If the goal is to create a canopy surface, then the Input Class Codes need to be

specified as 5 and the Input Return Values as first returns only. If the supplied lidar

data has only been classified as ground and unclassified, the Input Class Code is 1

for Unclassified, and the Input Return Value is first.

■ If the entire canopy is to be modeled, then the Input Class Codes need to be specified

as 3, 4, and 5 as these contain the upper, middle, and lower canopy returns,

respectively.

■ If the goal is to produce a ground surface, then the Input Class Codes need to be

specified as 2 and the Input Return Values as ANY_RETURNS.

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

The screen capture below represents ground returns using a series of points. In its raw

form, a multipoint feature class is not designed to be displayed; it is designed as an

efficient storage medium for the many millions of points found in a lidar dataset.

In this example, the points appear to be merging into a single dense mass of points, as

there are so many points contained in the feature class.

The ArcMap screen capture also shows the attribute table with the Shape field as a

Multipoint Z, the Intensity field is a BLOB, and the PointCount field shows how many

points are stored per record. The PointCount in this case is showing 3,500 records per

row. This is 3,500 points per record in the geodatabase. In a normal point feature class,

there is one point per record. Having many points per record enables the feature class to

be highly compressed, thus making the geodatabase an efficient method to store and

manage lidar data. The intensity value is a BLOB record. This means that each intensity

value is linked to each point via a specialized method in the Intensity field. To access this

BLOB field, the program needs to understand how to interpret the BLOB field.

January 2011 12

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

Figure 2

Ground Return Multipoint Feature Class

Ground return multipoint feature class shows Shape, Intensity, and PointCount fields.

Visualizing and

Storing Lidar Data

with ArcGIS

Visualizing Lidar

Data

Storing, mosaicking, and separating data in a multipoint feature class in a geodatabase is

the first stage of managing lidar data. The next stage is analyzing and visualizing the

data.

A forester or land manager may want to visualize the data to enable understanding of

■ Ground terrain

■ Canopy structure

■ Forest type and/or species

Esri White Paper 13

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

January 2011 14

A forester may want to analyze the data to enable understanding of

■ Tree heights

■ Vegetation biomass/density

■ Creek and river lines

■ Locations of road networks for both existing and new road planning

■ Existing terrains for the location of new plantations

When analyzing and visualizing the data, a decision needs to be made whether to convert

raw elevation point data to a geodatabase terrain or to an elevation raster file.

Although both these formats are useful for analysis in the forest application, selecting the

best format depends on the application.

Advantages of a

Raster

If the only source of data is lidar, then an elevation raster can be ideal as there is no

blending of additional data sources required. An elevation raster can be quickly produced

and created at any resolution. The raster often does not produce the highest-quality

results, but lidar data tends to be so dense that for many applications, the reduced

accuracy may be sufficient.

When working with lidar raster datasets, there will be situations where no returns are

recorded. With ground returns, this can be exaggerated where dense canopy exists and

the lidar cannot penetrate to the ground. Where no returns occur, holes or NULL values

will appear in the raster. This is the main disadvantage of using the Point To Raster tool.

These holes can be reduced with postprocessing techniques that will be discussed later in

this paper.

Advantages of a

Geodatabase Terrain

A geodatabase terrain is the optimal format to use if elevation data sources include lidar

(mass points); breaklines such as roads, water bodies, or rivers; and spot heights. The

geodatabase terrain is capable of blending these multiple data sources into one uniform

surface with a simple, easy-to-use wizard. A geodatabase terrain resides inside a feature

dataset in the geodatabase with the corresponding feature classes that were used to build it.

A geodatabase terrain references the original feature classes. It does not store a surface as

a raster or a triangulated irregular network (TIN). Rather, it organizes the data for fast

retrieval and derives a TIN surface on the fly based on the feature classes it resides with.

A terrain is not static and can be edited and updated as required. Local edits can be

performed, and rebuilding of the terrain is only required for the area of editing. If new

data is acquired, it can be easily added to the existing terrain and, again, only rebuilt for

the area of new data.

Supported data types for a terrain include

■ Points

■ Multipoints

■ Polylines

■ Polygons

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

Esri White Paper 15

In forestry applications, data types can be derived from

■ Bare earth lidar data

■ First/All return lidar data

■ Breaklines representing water body shorelines, rivers, culverts, and roads

Building and

Delivering DEMs

and DSMs from

Lidar

In forest applications, DEMs are useful for planning and operational activities. Terrain

beneath the tree canopy provides important information needed by silviculturists,

engineers, and equipment operators.

DSMs delineate aboveground vegetation and are therefore useful for understanding the

forest structure. They identify stands with similar characteristics, and when used in

conjunction with a DEM, use it to calculate tree heights.

Lidar data provides the user with the ability to make two distinct high-resolution

surfaces: a first return, or canopy surface, and a ground surface. Typically, the DSM will

contain tree canopy and buildings, and the DEM will contain bare earth or ground

returns. With the data loaded into a multipoint feature class in a geodatabase, it becomes

necessary to consider the workflow for DEM analysis.

Deciding whether to build a geodatabase terrain or a raster grid model will depend on the

requirements. Processing the data into a geodatabase terrain will be the most efficient

method for maintaining the data, but delivering to clients for consumption will require the

conversion of the final geodatabase terrain to a raster DEM format. This presents the user

with a problem—trying to process a single terrain into a single file with billions of points

will create a file too big to work with and process with most DEM applications. It is

therefore necessary to divide these raster files into smaller workable files. Again, the

problem presents itself of how these can then be served as a single terrain to clients.

Esri's ArcGIS Server Image extension solves this problem. It can consume the raster

DEM files and serve them through ArcGIS Server as an image service. The image service

can be consumed by ArcGIS clients as a visualized terrain or an elevation service. The

elevation service can then be utilized in ArcGIS extensions such as ArcGIS 3D Analyst

or ArcGIS Spatial Analyst for further terrain analysis.

The Workflow to

Create a Terrain and

Deliver to Clients

Typically, the workflow to get the data from the raw lidar files to a format that can be

consumed by client applications is as follows:

■ Convert the raw lidar data files to a multipoint feature class in a geodatabase.

■ Incorporate the multipoint feature class into a geodatabase terrain.

At this point, the terrain can be visualized and consumed by ArcGIS and geoprocessing

tools. If the datasets are very large, they can be served to GIS clients by the ArcGIS

Server Image extension. The workflow to move a geodatabase terrain to the image

service is to

■ Convert the geodatabase terrain to a series of DEM rasters.

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

■ Consume the multiple raster DEMs in the ArcGIS Server Image extension and serve

to clients as an elevation service or Web Coverage Service (WCS).

This workflow will be addressed in the ArcGIS Server Image extension section of this

paper.

Building a

Geodatabase Terrain

Geodatabase terrains reside in a feature dataset in the geodatabase. All features used by

the geodatabase terrain also reside in this feature dataset. This ensures that all the

contributing features have the same spatial reference. A terrain dataset is a

multiresolution, TIN-based surface built from measurements stored as features in a

geodatabase.

As all terrain datasets reside in a feature dataset in the geodatabase, the feature classes

used to construct the terrain must also reside in the feature dataset.

Here is how to generate a terrain dataset:

Initially, create a feature dataset in the geodatabase. From the File menu in ArcCatalog

™

select New > Feature Dataset.

It is important when creating a feature dataset that the coordinate system be the same as

the data to be used in the terrain dataset. All features that reside in a feature dataset,

including the terrain, have the same coordinate system.

With the feature dataset created, load the raw lidar data into a multipoint feature class in

the feature dataset using the LAS to Multipoint tool. If other elevation data sources are

available, such as breaklines and spot heights, copy them into the feature dataset.

Note: The feature class copy will fail if there is a mismatch between the coordinate

system of the feature dataset and the feature class.

January 2011 16

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

With all the source elevation data in the feature dataset, create the terrain in the feature

dataset. From the File menu in ArcCatalog, select New > Terrain. This can also be

achieved by right-clicking on the feature dataset and enabling the context menu.

Note: The ArcGIS 3D Analyst extension will need to be active to perform this operation.

This initializes the New Terrain wizard, which will lead you through the terrain

generation process.

The first form presented contains terrain characteristics. On this initial form

■ Specify the terrain name.

■ The feature dataset is scanned, and the wizard displays all feature classes that can

participate in the terrain. From the supplied list, select the multipoint feature class

containing lidar points with shape geometry and stream data to be used as breaklines.

■ Specify the average point spacing for the multipoint feature class. If the point

spacing is not known, the Point File Information geoprocessing tool can provide

point spacing for the supplied data files, as discussed earlier in this paper.

Esri White Paper 17

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

On the next wizard form, select feature class characteristics and decide how the feature

classes affect the interpretation of the surface and whether they are viewable at all

pyramid levels of the terrain. It is important to ensure that if there are breaklines or areas

of interest, the field containing the z-values is used in the height source.

January 2011 18

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

Terrain datasets can be pyramided using one of two point-thinning filters: z-tolerance and

window size. Pyramids are reduced resolution versions of the original underlying data,

which are displayed when working at small scales.

For DEM production, either method for pyramid production can be used. If rasterizing

from the full resolution point set, use the window size filter for terrain construction as it

is faster. If thinned data can be used for analysis, which is reasonable if the lidar is

oversampled for user needs, the z-tolerance filter can be used. Although more time

consuming, this method is most appropriate because it provides an estimate of vertical

accuracy of the thinned representation.

For DSM production, use the window size filter with the Z Max option. For DEM

production, use the window size filter with the Z Minimum option. It is not necessary to

perform secondary thinning unless the terrain is over relatively flat areas. In these

regions, performance will be improved by implementing secondary thinning.

Finally, generate the terrain pyramids.

Terrain pyramids are generated through the process of point reduction, also known as

point thinning. This reduces the number of measurements needed to represent a surface

for a given area. For each successive pyramid level, fewer measurements are used, and

the accuracy requirements necessary to display the surface drops accordingly. The

original source measurements are still used in coarser pyramids, but there are fewer of

them. No resampling or derivative data is used for pyramids. It takes time to produce

pyramids, so you need to consider how best to use them to your advantage.

Esri White Paper 19

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

Once the process is finished, the wizard will prompt the user to build the terrain. The

number of pyramids chosen to be built will dictate how long it will take to build the

terrain.

Below is an example of a terrain view. This example is a DEM derived from the level 2

classification in the raw lidar dataset.

January 2011 20

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

Figure 3

Terrain Derived from Ground Returns

A DEM or DSM can be built directly from the multipoint feature class using the Point to

Raster tool. The Point to Raster tool is ideal if the only data source is lidar data. The tool

takes a single feature class as input, is fast, and gives results suitable for most forestry

applications. As mentioned earlier, the disadvantage with this tool is that NULL value

cells will appear in the raster where no returns are encountered.

Building a Raster

DEM

The export formats supported by the Point to Raster tool include, but are not limited to,

.tif, .img, and Esri GRID.

The Point to Raster tool is initialized from the Conversion Tools toolbox

(ArcToolbox\Conversion Tools\To Raster\Point To Raster). On the form, select

■ The input multipoint feature class

■ The value field, which is the shape field from the multipoint feature class (This

contains the z-heights for the feature.)

■ The output raster (If the intended raster is an Esri GRID file, do not place an

extension on the file name. If the output is TIFF, terminate the file name with .tif.)

Esri White Paper 21

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

■ A cell assignment of MEAN or MAX (Set the MEAN values for an average surface

height. Setting the value to MAX is useful when producing a first return, canopy

surface.)

■ A cell size (This is important. Too fine, and the surface will have many NoData

cells; too coarse, and the surface will lose detail. A good rule here is four times the

average point spacing. In this example, the average point spacing is 0.6, so a cell size

of 3 meters is optimal [2.4 rounded up to the highest whole meter].)

The results from this tool are quick to generate, but as discussed, the frequency of the

NoData cells may make the raster DEM appear noisy. This problem can be further

magnified where vegetation cover is so dense that it has obscured the ground returns. The

diagram below shows the output from ground returns at a three-meter pixel size.

January 2011 22

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

Figure 4

Raster Returned from the Point to Raster Tool

It is possible to reduce this effect by postprocessing the raster DEM with a Python script

that incorporates the conditional function (con). When using the Conditional evalution

function, each cell in the raster DEM is evaluated for the NoData value. If the evaluation

is true, then a floating filter is used to gain the average values of the surrounding cells and

applied to the NoData cell. If the evaluation is false, the original raster is used. The

conditional (Con) expression looks like the following:

Con(<condition>, <true_expression>, <false_expression>)

Following is a sample Python statement for removing the NoData values:

from arcpy.sa import *

outputfile = Con(IsNull("point2ras"),

FocalStatistics("point2ras", NbrRectangle(3, 3, "CELL"),

"MEAN", "DATA"), "point2ras")

In this example, the condition applied is IsNULL.

■ If it returns true, then a FocalStatistics filter is applied. This finds the mean of the

values for each cell location on an input raster within a specified neighborhood and

sends it to the corresponding cell location on the output raster.

■ If it returns false, then the original cell value is returned.

Esri White Paper 23

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

Here is the result of such a filter.

Figure 5

Raster Returned after Postprocessing Using Conditional Evaluation Function

Analyzing Lidar

Data for Foresters

Calculating

Vegetation

Characteristics from

Lidar Data

With geoprocessing, lidar data can be made to reveal characteristics of a forest. The

forest height is calculated by analyzing the difference between the canopy surface and

ground surface. The vegetation density, or biomass, is calculated by analyzing the density

and frequency of returns for a given area. The following two sections outline these

processes.

Tree Height

Estimation

Tree height estimation is useful for growth analysis and approximating timber volume.

Areas of fast and slow growth can be quickly identified and fertilization schemes

developed based on these growth statistics.

With both the DEM and DSM generated from the lidar data, it is possible to estimate the

canopy height above the ground. To calculate the canopy height, simply subtract one

surface from the other by using the Minus tool found in the ArcGIS Spatial Analyst

toolbox (ArcToolbox\Spatial Analyst Tools\Math\Minus).

January 2011 24

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

The inputs for this tool are

■ Canopy surface (value 1)

■ Ground surface (value 2)

■ Output height raster

In the results below, blue portrays low vegetation, and red portrays high vegetation. In

this example, a break in tree growth can be clearly seen in the area highlighted as a hard

line between tall and low growth.

Figure 6

Height Estimation Raster

Esri White Paper 25

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

Biomass Density

Calculation

The biomass density will give an indication of tree vigor and growth. Where a forest is of

the same species, areas of low vigor can be quickly separated from areas of high vigor

and the causes investigated.

To calculate biomass density, it is necessary to have bare earth multipoints in one feature

class and all the aboveground points in another feature class. When creating the

aboveground feature class from the raw lidar data, it is necessary to include all vegetation

returns. According to the LAS 1.2 specification, vegetation classifications are 3, 4, and 5,

although some data suppliers may place vegetation in class 1 due to the expense of

classifying to the three vegetation classes. Ultimately, the lidar metadata will provide the

correct classification.

The key to determining biomass density is to calculate the raster file to be used with the

correct cell size. Normally, cell sizes four times the size of the average point spacing

should be used. This allows pixel averaging and removal of NULL cells. If smaller pixel

sizes are used, the frequency of the NULL cells increases and can bias the results. In the

examples used here, the average point spacing is 0.6, so the cell size is 3 meters

(2.4 rounded up to the nearest whole meter).

Point to Raster Tool

The first stage in determining biomass density is to convert the multipoint feature classes

to raster files. This is done with the Point to Raster tool. The Point to Raster tool is

initialized from the Conversion Tools toolbox (ArcToolbox\Conversion Tools\To

Raster\Point To Raster). When calculating density, the number of returns in a given area

is important rather than the elevation values returned, so the value field on the Point to

Raster form is irrelevant. As we are interested in the density of the returns, use the

COUNT qualifier for the cell assignment type. This provides an approximate density of

the lidar pulse returns.

Replacing NoData

Values as Zero

Vegetation Density

In the following two steps, the user takes all cells that have NULL values or cells of

NoDATA in them and assigns them a value of 0 to indicate the vegetation density is zero.

This is done so that all subsequent operations treat the NULLs as zeros (i.e, no vegetation

density) and real data values are returned.

January 2011 26

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

The first of the two steps uses the Is NULL tool in ArcGIS Spatial Analyst

(ArcToolbox\Spatial Analyst Tools\Math\Logical\Is Null). The tool designates all NULL

values in a raster file as zero. It reads the original grid file and writes out a binary file of 0

and 1. A 1 is assigned to values that are not NULL.

The second process then merges the original raster file with the NULL raster file so that

the resultant raster file has a complete range of values from 0 upward. There are no cells

that have an unassigned value. The tool used here is the Con tool in ArcGIS Spatial

Analyst (ArcToolbox\Spatial Analyst Tools\Conditional\Con).

When the Con tool is run, if a value of 0 is encountered, it is accepted as a true value. If a

value of 1 is encountered, the tool pulls the value from the original raster file. This results

in a final raster file without NULL values.

Repeat the above three processes for the aboveground or canopy raster.

Esri White Paper 27

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

Merging the

Aboveground and

Ground Results

Merge the aboveground density raster with the ground density raster to derive overall

density of returns. To do this, use the Plus tool in ArcGIS 3D Analyst

(ArcToolbox\Spatial Analyst Tools\Math\Plus).

Creating a Floating

Point Raster File

All rasters used have been integer value rasters. Each pixel in the raster is a whole

number. To calculate the density, use the Divide tool. The results from the Divide tool

range from zero to one. Using two integer rasters will result in an integer raster, which

will provide whole numbers for each cell and not a true representation of the result from

the Divide tool. To change the result of the output raster type, one of the input rasters

needs to be of the data type float. To transform a grid from data type integer to data type

float, use the Float tool in ArcGIS Spatial Analyst (ArcToolbox\Spatial Analyst

Tools\Math\Float).

Calculating Density

To calculate the density, use the Divide tool in ArcGIS Spatial Analyst

(ArcToolbox\Spatial Analyst Tools\Math\Divide). The result from the Divide tool is a

raster with a range between 0.0 and 1.0—hence the need earlier to create a float raster.

Dense canopy is represented by a value of 1.0, and no canopy is represented by a value of

0.0. In this case, the canopy returns are divided by the total returns.

January 2011 28

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

Note: Where data has been classified into ground (2) and unclassified (1), results may

vary from what is seen in this paper. The unclassified class may contain data other than

vegetation returns, which will influence the final result.

The canopy density raster is depicted in the screen capture below. Yellow represents low

density, and dark blue represents high density.

Figure 7

Canopy Density Raster

Esri White Paper 29

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

Distributing Large

Lidar Datasets

Lidar data is characterized by large volumes of data that can be stored in a geodatabase

terrain or a raster file. There are some situations where lidar data is so large that it should

not be represented by one raster file. Additionally, in some cases, multiple raster files

may also be slow to display, especially in situations where the geodatabase may have to

be transferred across a network.

In many situations, a mosaic dataset is the optimum solution for storing anad managing

large raster datasets. New at version 10, a mosaic raster dataset is a data model within the

geodatabase used to manage a collection of raster datasets (images) stored as a catalog

and viewed as a mosaicked image. Mosaic datasets have advanced raster querying

capabilities and processing functions and can also be used as a source for serving image

services through the ArcGIS Server Image extension. The ArcGIS Server Image

extension is used for delivering many large raster files to a client system quickly and

efficiently. It delivers images at only the current scale resolution and for the current view

extent.

Preparing Raster

DEM for Serving

with the ArcGIS

Server Image

Extension

As the ArcGIS Server Image extension only serves raster datasets, it is necessary to

output geodatabase terrain as raster files.

Use the Terrain to Raster tool (ArcToolbox\3D Analyst Tools\Conversion\From Terrain\

Terrain To Raster) to produce the rasterized version of the terrain. This tool provides

methods for interpolation of the terrain to grid cells, cell size, and the pyramid level to

use when producing the terrain. The extents for the output raster can be checked by

clicking the Environments button. If the output image is going to be used in further

analysis or inclusion in a mosaic dataset, Esri recommends the TIFF format. ArcGIS 10

includes support for the new BigTIFF format, which means TIFF images greater than

4 GB can be created.

For an interpolation method, natural neighbors provides the smoothest result. Although it

is not as fast to produce as linear interpolation, it is more accurate.

January 2011 30

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

Esri White Paper 31

Set the cell size to four times the density of the lidar data. This provides a smoothed

average. In this case, this is 0.6 meters, so 3 meters is a good size to work with (four

times the average spacing rounded up to the nearest whole meter). In forest applications,

there is no advantage to using a smaller size than this.

A significant concern here is the size of the terrain being exported. If the terrain is large,

the exported raster can exceed the maximum size of some raster formats. To overcome

this, a terrain can be divided into small resultant raster files. This is achieved by setting

the extents on the Environments tab. The extents can be defined by a group of features in

a feature class. These can then be used as input to a model that loops through the feature

class to export the final grid layers. The ArcGIS Server Image extension can then

consume and serve them as a seamless surface.

Serving an Elevation

Service through the

ArcGIS Server Image

Extension

The ArcGIS Server Image extension is used to serve and analyze mosaic datasets

seamlessly, such as aerial photographs and satellite images. It also has the capability to

serve terrain data through a specialized image service called an elevation service.

When the elevation service is serving raw elevation information, it can be consumed by

ArcGlobe

™

. It generates a surface to depict elevation over which other imagery can be

draped. It also can be used as an elevation input into 3D Analyst and Spatial Analyst

tools for complex models and terrain analysis.

A function can be applied to a mosaic dataset, which allows the elevation data to be

visualized. The function is applied to the mosaic dataset on the fly, reducing the need to

create multiple versions of the same dataset, thereby reducing dataset redundancy. Any of

the following functions can be applied to an elevation mosaic dataset and served through

the ArcGIS Server Image extension on the fly:

■ Hillshade

■ Shaded relief

■ Slope

■ Aspect

Typically, the best methods for visualizing the surface are hillshade and shaded relief.

Examples of these follow.

The advantage of using an elevation service is that an entire lidar dataset can be viewed

as a seamless image rather than a series of separate lidar files. This enables quick viewing

and interpretation of the entire DEM or DSM and performance of comparisons using

such tools as the Swipe tool or Flicker tool found on the Effects toolbar in ArcMap.

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

Figure 8

Example of a Hillshade

Figure 9

Example of a Shaded Relief

January 2011 32

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

Creating an

Elevation Mosaic

Dataset

Creating an elevation mosaic dataset in a geodatabase is the same process as developing a

standard image mosaic dataset, with the exception that it uses elevation sources rather

than standard image data. Although data accessed is raster data, it is interpreted as an

elevation source.

The process outlined below is used to generate an elevation mosaic dataset. It uses the

ArcGIS 10 mosaic dataset geoprocessing tools. The Mosaic Dataset tools are located in

ArcToolbox\Data Management Tools\Raster\Mosaic Dataset.

To create a new mosaic dataset

■ Select the Create Mosaic Dataset geoprocessing tool. The Create Mosaic Dataset tool

has three primary entry fields:

● Output Location—This is the geodatabase that will contain the mosaic dataset.

● Mosaic Dataset Name

● Coordinate System—This is the coordinate system of the mosaic dataset, which

may differ from the underlying raster sources.

There is no need to specify the type of service, as this is interpreted from the pixel

source when the raster data is added to the mosaic dataset.

■ Select OK; the mosaic dataset will be created.

The mosaic dataset can be considered a container to which the raster data now needs to

be added.

To add data to the mosaic dataset

■ Select the geoprocessing tool Add Rasters To Mosaic Dataset. Input to this form can

be divided into two distinct groups—information about storage of the rasters and

advanced information about the rasters.

Esri White Paper 33

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

The first group of inputs includes

● Selecting the mosaic dataset created earlier

● The Raster Type drop-down list—This drop-down list can specify different

image sources/sensors such as Landsat or QuickBird. In this sample set, the

elevation source is a 32-bit .tiff file, so Raster Dataset is specified.

● The Input for this raster source is Workspace. All the exported terrain data files

from earlier processing are stored in a folder, which here is referred to as the

Workspace.

● Update Overviews—This generates the pyramids over the entire mosaic dataset.

The mosaic dataset considers the entire raster set as a seamless mosaic and

builds lower-resolution pyramid files over the single seamless mosaic.

January 2011 34

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

The second group of inputs is selected from Advanced Options. This group of inputs

describes the characteristics of the rasters and how to work with them. Generally, the

following settings are made:

● The coordinate system of the source rasters

● The option to recursively search subfolders if the data resides in more than one

folder

● The option to include or exclude duplicate rasters or overwrite existing rasters

This option can be useful if some raster data is found to be corrupt and needs to

be updated.

● Build Raster Pyramids option for each of the source rasters if they do not

already exist

Selecting this option increases the performance of the display of the raster

dataset when served through the ArcGIS Server Image extension.

● Calculate Statistics option to provide feedback to the system of each raster's

pixel range

Esri White Paper 35

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

■ Select OK to populate the mosaic dataset with select raster datasets and generate the

overviews of the mosaic dataset.

The resultant mosaic dataset can now be loaded into ArcMap. A mosaic dataset is

composed of three principle components:

● A Boundary layer, which defines the extent of the mosaic dataset

● A raster Footprint layer, which is the extent for each of the rasters in the mosaic

dataset

● The Image layer, which represents the entire image layer being displayed at the

required resolution for the current display scale

Figure 10

Elevation Mosaic Dataset

In figure 10, the Footprint layer is composed of the original raster data sources and the

generated overviews. The overviews are smaller; lower-resolution images based on the

original raster data provide the mosaic dataset with its performance.

This mosaic dataset can now be published as an image service using the ArcGIS Server

Image extension. To publish a mosaic dataset, right-click on the mosaic dataset from

within ArcCatalog and select Publish to ArcGIS Server.

January 2011 36

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

Publishing the mosaic dataset is the final stage in the elevation service creation. The

elevation service can be used as an elevation source for terrain visualization in ArcGlobe

or as input to ArcGIS 3D Analyst or ArcGIS Spatial Analyst geoprocessing tools such as

contour or viewshed.

Visualizing an

Elevation Service

The published elevation service is useful to gain an understanding of the terrain but often

does not provide the visual context if the service is to be used as a background. The next

section shows how to apply visualization to the mosaic data source through the use of a

mosaic dataset function. Mosaic dataset functions are applied to the raster data on the fly

as the mosaicked image is accessed and viewed.

While a function can be applied to a mosaic dataset directly, it is often not desirable, as it

affects the final output and changes the values returned from the primary mosaic dataset.

A referenced mosaic dataset can be created that references all the original parameters

from the primary mosaic. A unique function can then be applied to the referenced mosaic

dataset, which provides a different visualization of the lidar data from the source mosaic

dataset, and the visualization is created on the fly for the area of interest. Multiple

referenced mosaic datasets can be created for the original source mosaic dataset.

In this example, a referenced mosaic dataset is created, and the function used is the

Shaded Relief function.

From within ArcCatalog, perform the following steps to apply a Shaded Relief function

to a referenced mosaic dataset.

Esri White Paper 37

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

Creating a

Referenced Mosaic

Dataset

A referenced mosaic dataset is created by launching the Create Referenced Mosaic

Dataset geoprocessing tool from the Mosaic Dataset toolbox found at ArcToolbox\Data

Management Tools\Raster\Mosaic Dataset.

When the dialog box appears, enter

■ The Input Mosaic Dataset as the Ground Mosaic created earlier

■ The Output Mosaic Dataset as the name for the New Mosaic Dataset

The Coordinate System should be automatically populated from the Input Mosaic. Check

the Build Boundary option, then select OK to create the referenced mosaic dataset.

Before the function can be applied to the referenced mosaic dataset, the statistics of the

dataset need to be calculated.

To calculate the statistics

■ Select the referenced mosaic dataset and right-click to enable the context menu.

January 2011 38

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

■ Select the Calculate Statistics option.

The Calculate Statistics dialog box appears.

■ Select OK to have the statistics calculated.

Esri White Paper 39

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

Applying a Mosaic

Function

With the referenced mosaic dataset created, a function now needs to be added to the

mosaic dataset.

To add the Shaded Relief function

■ Right-click the referenced mosaic dataset and select Properties.

● Select the Functions tab on the Mosaic Dataset Properties dialog box. The

Functions tab is where the functions are configured for the mosaic dataset. One

or many functions can be applied to the mosaic dataset to provide on-the-fly

interpretations of the original mosaic dataset.

● Right-click Mosaic Function; the functions menu appears.

January 2011 40

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

● Select Insert > Shaded Relief Function.

The parameters for the Shaded Relief form appear. In this example, the yellow

to dark red color ramp is used, and the default settings are used for Azimuth,

Altitude, and Z Factor.

Esri White Paper 41

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

● Select OK; the Shaded Relief function is added to the Function Chain.

● Now, select OK to apply the function to the mosaic dataset.

The resultant visualization in figure 11 is of the referenced mosaic dataset of the ground

returns.

Figure 11

Visualized Shaded Relief Mosaic Dataset

This mosaic dataset can now be published as an image service using the ArcGIS Server

Image extension. As before, to publish a mosaic dataset, right-click on the referenced

mosaic dataset from within ArcCatalog and select Publish to ArcGIS Server.

Publishing the referenced mosaic dataset is the final stage in the visualized elevation

service creation. It can now be consumed in ArcMap and ArcGlobe and published on the

Web through any one of Esri's Web APIs.

January 2011 42

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

Esri White Paper 43

Estimating Tree

Height Using Mosaic

Dataset Functions

As was detailed earlier in this paper, it is possible to estimate the height of vegetation

from lidar by performing algebraic operations between the DSM and DEM. When

using the mosaic datasets, functions can be applied that return the same results. The

advantage of using the mosaic datasets is that there is no need to create a height

estimation raster, as the Mosaic Dataset function creates the height estimation on the fly

for the area of interest on the screen, removing the need to create extra raster datasets.

The creation of a height estimation mosaic dataset requires two elevation mosaic datasets

as the source inputs: a DEM mosaic dataset and a DSM mosaic dataset. The authoring of

the DEM mosaic dataset was described in the section Creating an Elevation Mosaic

Dataset. To create a DSM service, follow this same method but use the DSM terrain as

the source data.

The height estimation mosaic dataset is a referenced mosaic dataset that takes the input of

the two mosaic datasets and performs an arithmetic function between them to produce a

third service.

Creating the Height

Estimation Mosaic

Dataset

When creating the height estimation mosaic dataset, it is necessary to create a referenced

mosaic dataset and add the two source mosaic datasets files as two distinct steps.

The referenced mosaic dataset is created by launching the Create Referenced Mosaic

Dataset geoprocessing tool from the Mosaic Dataset toolbox found at ArcToolbox\Data

Management Tools\Raster\Mosaic Dataset.

■ Select Input Mosaic Dataset as the mosaic dataset that displays the canopy data. It is

important that the canopy be selected for the input mosaic dataset, as the function

applied later will be against the input mosaic dataset.

■ Select Output Mosaic Dataset as the name for the new mosaic dataset.

■ Specify the Coordinate System for the mosaic dataset.

■ Check the Build Boundary option.

Under Pixel Properties,

■ In Number of Bands, enter 1.

■ In Pixel Type, choose 32_BIT_FLOAT.

This is the same as the source mosaic datasets.

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

■ Select OK to create the referenced mosaic dataset.

The referenced mosaic dataset is created, and the arithmetic function now needs to be

added.

Applying the

Arithmetic Function

To add the arithmetic function

■ Right-click on the referenced mosaic dataset and Select properties.

January 2011 44

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

■ Select the Functions tab on the Mosaic Dataset Properties dialog box.

■ Right-click Mosaic Function and select Insert > Arithmetic Function.

■ On the Raster Function Properties dialog box, make the following settings:

● Input Raster 1 is left untouched, as this is the canopy mosaic dataset specified

earlier.

● In Input Raster 2, click the Browse button, then choose Ground Mosaic Dataset.

● Operation is set to Minus. This is how Raster 2 interacts with Raster 1.

Esri White Paper 45

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

■ Select OK; the function will be applied to the mosaic dataset. As can be seen in the

screen shot below, the ground mosaic now forms part of the Arithmetic Function.

■ Select OK on the Mosaic Dataset Properties dialog box; the function will be applied

to the referenced mosaic dataset.

For ArcMap to render the mosaic correctly, the statistics for the referenced mosaic

dataset need to be calculated.

To calculate the statistics

■ Select the referenced mosaic dataset in ArcCatalog and right-click to enable the

context menu.

■ Select the Calculate Statistics option. The Calculate Statistics dialog box appears.

■ Select OK to calculate the statistics.

The referenced mosaic dataset can now be added to ArcMap.

Note: By default, the height estimation mosaic dataset is in grayscale. To add context to

the raster, a color ramp is applied. In figure 12, the color ramp is yellow to green to dark

blue.

January 2011 46

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

Figure 12

Height Estimation Image Service

Figure 12 shows the results of the height estimation mosaic dataset. In this example, light

green represents tall vegetation and dark blue areas represent low vegetation.

The real vegetation height values are returned as interpreted heights from the lidar data.

Thus, any pixel can be interrogated to return the tree height. The results can be used by

the foresters to aid in forest stand delineation and field inventory layout and to estimate

size class and other forest inventory metrics.

This referenced mosaic dataset can now be published as an image service using the

ArcGIS Server Image extension. To publish a mosaic dataset, right-click on the mosaic

dataset from within ArcCatalog and select Publish to ArcGIS Server.

It is important to note here that the results displayed by the ArcGIS Server Image

extension are only calculated on the fly for the extent of the ArcMap window and at the

display scale of the map window. This means that broad acreage data sources can be used

as the input for the elevation services, and the results are only calculated for the screen

area. This reduces data duplication and returns results quickly.

Esri White Paper 47

Lidar Analysis in ArcGIS 10 for Forestry Applications

J-9999

January 2011 48

Conclusion

This paper has demonstrated that the benefits to the forest industry for the use of lidar are

wide and varied. These include methods to

■ Analyze and validate raw lidar data with ArcGIS before any extensive processing

occurs.

■ Store and manage millions of lidar returns within the geodatabase in a seamless

dataset, regardless of the number of original lidar files.

■ Extract DEMs and DSMs from the lidar data and store them as terrains in a

geodatabase or as raster elevation files.

■ Extract vegetation density estimates and tree height estimates from lidar, which aid

in growth analysis, fertilization regimes, logging operations, and supplementing and

validating traditional field forest mensuration techniques.

■ Serve and analyze large amounts of lidar data as a seamless dataset to GIS clients,

reducing the need to analyze each lidar file on a file-by-file basis, providing good

overall analysis of the forest.

ArcGIS is a complete system for managing, storing, and analyzing lidar data. Coupling

ArcGIS with the ArcGIS Server Image extension enables organizations to distribute and

access large amounts of lidar data quickly and efficiently without the need to produce

additional resultant datasets.

Acknowledgments

The author would like to thank Forestry Tasmania, Australia, for the sample lidar data

used as examples throughout this document.

Printed in USA

About Esri

Since 1969, Esri has been helping

organizations map and model our

world. Esri’s GIS software tools

and methodologies enable these

organizations to effectively analyze

and manage their geographic

information and make better

decisions. They are supported by our

experienced and knowledgeable staff

and extensive network of business

partners and international distributors.

A full-service GIS company, Esri

supports the implementation of GIS

technology on desktops, servers,

online services, and mobile devices.

These GIS solutions are exible,

customizable, and easy to use.

Our Focus

Esri software is used by hundreds

of thousands of organizations that

apply GIS to solve problems and

make our world a better place to

live. We pay close attention to our

users to ensure they have the best

tools possible to accomplish their

missions. A comprehensive suite of

training options offered worldwide

helps our users fully leverage their

GIS applications.

Esri is a socially conscious business,

actively supporting organizations

involved in education, conservation,

sustainable development, and

humanitarian affairs.

Contact Esri

1-800-GIS-XPRT (1-800-447-9778)

Phone: 909-793-2853

Fax: 909-793-5953

info@esri.com

esri.c om

Ofces worldwide

esri.com/locations

380 New York Street

Redlands, CA 92373-8100 USA