02

Chapter

Operational Guide
to Tender a Digital
Elevation Model
Digital Elevation Models: Design Issues for Consideration
In planning DEM design and implementation, project managers need to identify a combination of data sources,
collection methods and modalities that will get them to a cost-effective response that meets the user’s needs.
There is, however, no simple template, as any particular DEM application will have its own character and
circumstances. This section aims to provide broad guidance for meeting project requirements and operational
parameters, all within a specified budget, weighing various constraints against different modalities, collection
methods, and data sources. A specialist can provide advice on the technical specifications to be defined in the
Scope of Work/Terms of Reference. It is therefore recommended that a DEM specialist be hired on projects that
require the acquisition and use of DEMs to avoid wasting large amounts of funds on generating/acquiring DEMs
that do not meet the needs of the project. The specialist can also provide guidance on whether the delivered
product meets the specifications described in the ToR.

Decision Points to Acquire a DEM for Projects

Need for Digital Elevation Model (DEM) identified for a project.

Hire DEM technical specialist

Plan for data storage and data sharing of the generated DEM

Define the ideal technical spec required for the DEM, based on the intended use

Define the geographical Area Of Interest (AOI) where the DEM is needed

For the AOI, search for:

• Existing DEM datasets generated for other projects covering the same area
• Investigate if globally available off-the-shelf DEMs serve the purpose, for example, WorldDEM (http://www.geo-
airbusds.com/worlddem/), ALOS World 3D (http://alos-world3d.jp/en/), NEXTMAP http://www.intermap.com/
data/nextmap-world-30)

19
• Look into the possibility of using existing raw datasets of the area that can be used to derive a DEM, such as
stereo aerial photography.
20 Digital Elevation Models – A Guidance Note on How Digital Elevation Models Are Created and Used

• Research any on-going projects that are planning on generating a DEM for the AOI. (World Bank funded
projects or externally funded projects)—if so, does the timing of the data generation suit the project timeline?

If there are existing DEMs that can be used for the project, purchase the DEM as goods.

If there are no existing DEM datasets that can be used for the project, commission a new DEM dataset for the AOI.

In most cases, this will involve doing a Lidar survey. Lidar surveys can generate very high-resolution DEM. In
cases where a large area is to be covered, it may make sense to compromise on the resolution and generate a
DEM with lower resolution by acquiring satellite radar data (Interferometric Synthetic Aperture Radar- IfSAR) or
use other means.

After all the technical requirements are set following discussions with the client and technical consultant, issue
an EOI, Limited International Bidding (LIB) where only vendors with good track records are invited to bid.

Requirements and Options

Project requirements are driven by the information needs of the end user. Several of them—vertical accuracy,
spatial resolution, study area size, and study area location—effectively predetermine the options for data
acquisition. Realizing DEM requirements can be further constrained by other factors influencing data collection,
such as time allowed for delivery, prevailing weather conditions in the study area, vegetation and structures, and
last but not least, available budget.

Many people will associate the process of generating DEMs with Lidar surveys; however, Lidar is not the only
means to generate a DEM. A DEM with the same spec can be generated using various mediums and datasets.
These different mediums vary from satellite sensors, sensors attached to aircrafts as well as traditional surveying
methods on the ground. Technical details of the different mediums can be found in the accompanying document
“Technical Annex: How DEMs are created: a brief introduction to remote sensing modalities.”

In this section, the technical specifications that need to be specified for a DEM to be generated and some
decision making criteria when there are more than one options will be described. Once the specs for the output
DEMs are defined, it is up to the vendor to suggest the most economic and efficient way to generate the required
DEM, taking into account the context of the overall project.

Vertical Accuracy
Accuracy (especially vertical) is the primary quality metric for DEM products. Accuracy expressed as vertical
error can be characterized as Very High (<0.5m), High (0.5m to 1.0 m), Medium (1.0m to 5.0m), Low (5.0m to
10.0m), or Very Low (>10.0m). Data from airborne collections with ground control usually have the best accuracy
but do not generally have the large area coverage available from archived satellite data. Inaccessible terrain may
make ground control difficult and reduce the attainable accuracy. Higher accuracy requires photogrammetry or
Lidar; lower accuracy allows use of IfSAR and satellite archive data.

The Spatial Analysis Group at the University of Southern Queensland (Australia) has summarized key
requirements (vertical accuracy) for a variety of applications as shown in Table 1. The applications have been
grouped in ascending order of accuracy requirements.
 Operational Guide to Tender a Digital Elevation Model 21

Table 1. Key accuracy requirements for a range of application areas.

Grouping Application Area Accuracy

Very High Farm layout redesign – cultivation 0.05–0.5m
<0.5m Hydrological modeling (incl. floodplain) <0.5m
Land and water management plans <0.5m
Insurance risk and assessment <0.5m
High Disaster response 0.5–1m
0.5–1.0m Infrastructure planning and risk assessment 0.5–1m
Water management plans 0.5–1m
Cross slope/batter analysis <1m
Planning scheme/development assessment +/– 1m
Transport corridor planning +/– 1m
Soil erosion control and modeling +/– 1m
Bio-security—disease spread, spray drift +/– 1m
Disaster planning and management (except flood) +/– 1m
Medium Riparian management +/–1–2m
1.0–5.0m Risk management +/–1–2m
Noise studies/assessment—corridor planning +/– 1–2m
Telecommunications planning, visibility analysis 1–5m
Salinity prediction and control +/– 5m
Low Visibility analysis — tourism 5–10m
5.0–10.0m Environmental impact assessment and management +/– 5–10m
Natural resource management 5–10m
Very Low >10m Tourism 20m
Source: Courtesy of the Spatial Analysis Group at the University of Southern Queensland, Australia.

Spatial Resolution (ground sample distance or point spacing)

Spatial (horizontal) resolution is driven by the level of detail and point spacing in the model desired by the
user. The value should be less than the size of the smallest terrain features the user desires to be represented.
For instance if buildings are to be captured, the point spacing should be smaller than the dimensions of the
smallest buildings in the area of interest. Spatial resolution can be characterized as high (<1m), medium (1m to
5m), or low (>5m). Higher resolution is generally needed for areas with rapid changes in elevation, such as steep
terrain or urban areas, while smooth terrain can be suitably covered at lower resolution. Photogrammetric or
Lidar systems are best for higher resolution applications on the order of <1.0m. Medium or lower accuracy
applications allow use of IfSAR, on the order of 1m to 5m, and satellite archive data.

Coverage Location/Area
Coverage extent, accuracy and resolution are typically the largest drivers of project cost and schedule. Study
location drives deployment costs for aircraft and field support, and can be a major cost factor for custom
collections. Inaccessible terrain can make ground control difficult and reduce the attainable accuracy. The
extent of the area is an obvious cost factor, especially for aircraft collections, with costs quoted on an area basis
(e.g. $/km2). Due to the vagaries of large-scale projects, providers of aerial data tend to require longer schedules
to reduce risk. In contrast, archived satellite data, provided it has the needed resolution and accuracy, has the
22 Digital Elevation Models – A Guidance Note on How Digital Elevation Models Are Created and Used

lowest schedule risk for large area projects. From a cost perspective, custom airborne photogrammetric or
Lidar are good for smaller areas. IfSAR and satellite archive data are best for large areas if accuracy and
resolution permit.

Weather Constraints
Some areas of the world have significant seasonal weather activity that may render some modalities
ineffective—especially photogrammetry and Lidar—and limit the time window for data collection. IfSAR is
essentially immune to weather, making that modality attractive for areas that are chronically cloud covered. Smoke
is a special case that may be encountered in disaster response applications. Smoke obscuration is problematic for
photographic systems and to a certain extent Lidar, but the operating wavelengths of IfSAR systems are immune
to smoke effects. Heavy haze encountered in some developing urban areas may also present a challenge to
photographic approaches, but can be overcome by Lidar and IfSAR. IfSAR is best for chronically cloudy or foggy
areas, and good for haze conditions and smoke environments. Lidar is also good for heavy haze conditions.

Timeline Constraints
Product delivery timelines, if specified, can impact the options available to the program manager. Products with
the shortest timeline will be those available from archives, though that data may not satisfy requirements of
accuracy, coverage, or temporally relevant scene content. Tight timelines may result in higher procurement costs.
Use of archive data has the shortest timeline if product quality is acceptable.

Foliage Constraints
Foliage can obscure the ground from above, preventing accurate determination of a bare-earth DEM, especially
for photogrammetric methods. Lidar sensors are capable of foliage penetration provided the point density
(spatial resolution) is sufficiently high (small spacing) to get enough pulses between the leaves to the ground.
This will typically require closer flight line spacings, with a consequently higher operational cost. Most
common IfSAR products are based on higher frequency systems such as X-band and thus have poor vegetation
penetration. However, longer wavelength IfSAR, especially P-band, has excellent foliage penetration capability.
Photogrammetry is poor for determination of a bare-earth DEM under heavy foliage. Lidar can be acceptable
with sufficiently high point density. Longer wavelength IfSAR (P-band) has good foliage penetration.

Urban Buildings and Steep Terrain Constraints

Steep terrain, buildings, and other tall structures can create gaps in coverage due to shadowing of Lidar or
IfSAR sensors. This can be overcome for Lidar (and photogrammetry) by closer spacing of flight lines, with
consequently higher operational costs. IfSAR can be more difficult as the system is less flexible. IfSAR is less
suitable for tall urban structures or very steep terrain. Photogrammetry and Lidar can be used with more closely
spaced flight lines.

Budget Constraints
Budget limitations can pose a major constraint on what acceptable options are available for a given project.
Costs for archive data are generally lower on an area basis (cost/km2). The lower cost is usually accompanied by
strict limitations on data use, as the archive owner typically retains ownership of the data and provides a limited
use license to the project. Costs for specific applications may vary significantly, in particular for those requiring
custom data collections. For example, the mobilization costs associated with deployment of a vendor’s aircraft
and equipment to the study area may add €10,000–50,000 (approximately US$13,000–67,000) to the contract
cost depending on distance travelled. The most predictable costs come from archived satellite data sources such
as Digital Globe and Astrium, which have well-established catalog prices. Table 2 presents generalized data on
DEM costs for various vendors and remote sensing modalities.
 Operational Guide to Tender a Digital Elevation Model 23

Table 2. DEM product costs for various remote sensing modalities and vendors

Vertical Price $/km2

Vendor DEM Product Accuracy (approx.) Licensing
Various Photogrammetry (Air) 0.25–0.3m $440 User owned
Various Lidar (Air) 0.15–0.3m $180 User owned
Fugro IfSAR Dual X/P Band (Air) 2–5m $80 User owned
DigitalGlobe Photogrammetry (Space) 4m $32 Limited license
Astrium IfSAR “Elevation10 Package” (Space) 5–10m $50 Limited license

IfSAR and space photogrammetry are generally much less expensive but with less vertical accuracy. Archive data
is lowest cost but usually has limited rights for dissemination. Custom airborne photogrammetry and Lidar are
the most expensive but provide the best accuracy and least dissemination restrictions.

Suitability Matrix
To assist decision makers in defining an appropriate remote sensing response for various applications, Table 3 shows
a “Suitability Matrix” which compares the various data acquisition options for a manager to a range of requirements.

Table 3. A suitability matrix that can be used to map a remote sensing modality to different project environments,
time scales, resolution requirements, and cost

Modality Method Source

Photo Lidar IfSAR Air Space Custom Archive Comments
Accuracy
Very/High X X X X High accuracy is best achieved by low flying
Accuracy aircraft with Lidar or photographic sensors.
< 1.0 m Archives tend to have lower accuracy data
compared to custom collects.
Medium X X X X Available from aircraft and some high
Accuracy performance satellite systems. Archives
1.0 to 5.0 m may have contain this quality.
Very/Low X X X Suitable satellite archive info is available for
Accuracy most areas of the world.
> 5.0 m
Spatial Resolution
High X X X X High resolution is best achieved by low
Resolution flying aircraft with lidar or photo sensors.
< 1.0 m
Medium X X X X Airborne IfSAR data can compete effectively
Resolution with photo and Lidar in this domain.
1.0 to 5.0 m
Low X X X Space IfSAR is excellent for low resolution
Resolution DEMs, satellite sensors are a great source
> 5.0 m of low resolution data at reasonable cost.
(continued on next page)