Draft USGS Report

DRAFT – DO NOT CITE

Secondary Identification and /or Statement of Cooperation

The Users, Uses, and Value of Landsat and Other
Moderate-Resolution Satellite Imagery in the United
States: Executive Report
By Holly M. Miller, Natalie R. Sexton, Lynne Koontz, John Loomis, Stephen R. Koontz, Catherine M.
Lundy, and Caroline Hermans

Report Series 2010–XXXX

U.S. Department of the Interior
U.S. Geological Survey

U.S. Department of the Interior
1

DRAFT – DO NOT CITE

KEN SALAZAR, Secretary
U.S. Geological Survey
Marcia McNutt, Director
U.S. Geological Survey, Reston, Virginia 2010

For product and ordering information:
World Wide Web: http://www.usgs.gov/pubprod
Telephone: 1-888-ASK-USGS

For more information on the USGS—the Federal source for science about the Earth,
its natural and living resources, natural hazards, and the environment:
World Wide Web: http://www.usgs.gov
Telephone: 1-888-ASK-USGS

Suggested citation:
Miller, H.M., Sexton, N.R., Koontz, L., Loomis, J., Koontz, S.R., Lundy, C.M., and Hermans, C., 2010,
The users, uses, and value of Landsat and other moderate-resolution satellite imagery in the
United States—Executive report: Fort Collins, CO, U.S. Geological Survey Open-File Report 2010XXX, XX p.
Any use of trade, product, or firm names is for descriptive purposes only and does not imply
endorsement by the U.S. Government.
Although this report is in the public domain, permission must be secured from the individual
copyright owners to reproduce any copyrighted material contained within this report.

2

DRAFT – DO NOT CITE

Contents
Contents ................................................................................................................................................................... 3 
Figures ...................................................................................................................................................................... 3 
Tables ....................................................................................................................................................................... 4 
Acknowledgments .................................................................................................................................................. 5 
Introduction ............................................................................................................................................................. 6 
User Identification .................................................................................................................................................. 7 
User Survey ............................................................................................................................................................. 8 
Results ...................................................................................................................................................................... 8 
Analysis ................................................................................................................................................................ 8 
Statistical Significance and Interpretation .................................................................................................... 9 
Diversity of the Sample ...................................................................................................................................... 9 
User Types ......................................................................................................................................................... 10 
Demographics ................................................................................................................................................... 10 
Use of Landsat Imagery ................................................................................................................................... 10 
Types of Imagery........................................................................................................................................... 10 
Scales and Locations of Projects .............................................................................................................. 11 
Application Areas ......................................................................................................................................... 11 
How Imagery is Used ................................................................................................................................... 12 
Level of Landsat Use in Work ..................................................................................................................... 12 
“Local Users” ................................................................................................................................................ 13 
Discussion: Use of Landsat Imagery ......................................................................................................... 13 
Changes in Use of Landsat Over Time .......................................................................................................... 13 
Discussion: Change in Use of Landsat Over Time .................................................................................. 14 
Impacts of No Cost Data Policy ..................................................................................................................... 14 
Discussion: Impacts of No Cost Data Policy............................................................................................ 15 
Value of Landsat Imagery................................................................................................................................ 15 
Importance and Satisfaction ...................................................................................................................... 16 
Benefits of Landsat ....................................................................................................................................... 16 
If Landsat Was No Longer Available… .................................................................................................... 17 
Willingness to Pay for Landsat Imagery ................................................................................................... 18 
Single-Bounded WTP Analysis............................................................................................................... 19 
Double-Bounded WTP Analysis ............................................................................................................. 20 
Discussion: Single-Bounded Vs. Double-Bounded WTP Analyses ................................................. 20 
Conclusion ............................................................................................................................................................. 21 
References Cited .................................................................................................................................................. 22 

Figures
Figure 1.Types of imagery users. ...................................................................................................................... 24 
Figure 2. Sectors of respondents who are currently using Landsat. .................................................................. 25 

3

DRAFT – DO NOT CITE

Figure 3.Locations of projects using Landsat imagery in year previous to survey. ................................ 26 
Figure 4.Scales of projects using Landsat imagery in year previous to survey. ..................................... 27 
Figure 5.Applications of Landsat imagery in year previous to survey. ...................................................... 28 
Figure 6.General use of moderate-resolution imagery among respondents who are currently using
Landsat. ................................................................................................................................................................. 29 
Figure 7.Level of use of Landsat imagery in work in year previous to survey. ........................................ 30 
Figure 8.Change in use of Landsat imagery in past 10 years and next 5 years. ...................................... 31 
Figure 9.Potential use of Landsat imagery among past Landsat users. .................................................... 32 
Figure 10.Reactions to SLC-off on Landsat 7. ................................................................................................. 33 
Figure 11.Sources of Landsat imagery acquisitions before and after the imagery became available at
no cost (2008 versus 2009). ................................................................................................................................ 34 
Figure 12.Patterns of acquisitions of Landsat imagery from EROS before and after it became
available at no cost (2008 versus 2009). .......................................................................................................... 35 
Figure 13.Importance of Landsat imagery to the work of respondents..................................................... 36 
Figure 14.Importance of and satisfaction with certain attributes of Landsat. ......................................... 37 
Figure 15.Preferred imagery versus imagery most likely to acquire within budget constraints among
respondents who would substitute other imagery for Landsat if it was no longer available. .............. 38 
Figure 16.Preferred imagery versus imagery most likely to acquire within budget constraints among
respondents who would choose a different imagery based on budget constraints. ............................. 39 
Figure 17.Single-bounded demand curve of willingness to pay for imagery to replace Landsat. ....... 40 
Figure 18.Single-bounded logit model results for willingness to pay for imagery to replace Landsat.
................................................................................................................................................................................. 41 
Figure 19.Single- and double-bounded demand curves of willingness to pay for imagery to replace
Landsat. ................................................................................................................................................................. 42 

Tables
Table 1.Mean percent of different types of imagery used in the year previous to the survey among
respondents who used a mix of moderate-resolution imagery. ................................................................. 43 
Table 2.Applications of Landsat imagery. ....................................................................................................... 44 
Table 3.Acquisitions of Landsat imagery before and after imagery became available at no costs
(2008 versus 2009). ............................................................................................................................................... 45 
Table 4.Percentages of respondents who would take each of three actions if Landsat imagery was
no longer available. ............................................................................................................................................. 45 
Table 5.Percentages of respondents who would use each of three types of information as
substitutes for Landsat imagery if it was no longer available. .................................................................... 45 

4

DRAFT – DO NOT CITE

Acknowledgments
This study was funded by the U.S. Geological Survey’s (USGS) Land Remote Sensing
Program and is part of a larger study being conducted in conjunction with the USGS Western
Geographic Science Center (WGSC) in Menlo Park, Calif. We would like to thank: Eric Wood,
Tom Loveland, and John Dwyer from the USGS Earth Resources Observation and Science (EROS)
Center in Sioux Falls, S.D., for their invaluable help in constructing, reviewing and testing the
survey instrument; Katie Walters (FORT), Makiko Hong and Tamara Wilson (WGSC) for their
work during the sampling phase of the project; those at EROS and South Dakota State University
who participated in a pre-test of the survey; and the numerous people who reviewed the survey and
provided constructive feedback, including Richard Bernkopf (WGSC), Molly Macauley (Resources
for the Future), those who attended our session at the 17th William T. Pecora Memorial Remote
Sensing Symposium, and members of the Landsat Science Team.

5

DRAFT – DO NOT CITE

The Users, Uses, and Value of Landsat and Other
Moderate-Resolution Satellite Imagery in the United
States: Executive Report
By Holly M. Miller 1 , Natalie R. Sexton, Lynne Koontz, John Loomis, Stephen R. Koontz, Catherine M.
Lundy, and Caroline Hermans

Introduction
A variety of satellites provide remotely sensed images of the earth at different resolutions,
generally categorized as high, moderate, or low resolution. According to the 2007 Future of Land
Imaging Interagency Working Group report (FLIIWG, 2007), moderate-resolution imagery (MRI)
is defined as imagery which:
covers relatively large geographic areas per scene (>60 km2),
has a spatial resolution between 5 and 120 meters,
is characterized by repetitive coverage, and
includes measurements from several portions of the electromagnetic spectrum.
MRI is provided by an assortment of satellites operated by governments and private businesses
around the world. One main source of MRI is the Landsat satellites operated by the U.S. Geological
Survey (USGS) who receives, processes, distributes, and archives Landsat data at the Earth
Resources Observation and Science (EROS) Center. Currently, there are two Landsat satellites
orbiting earth: Landsat 5 and Landsat 7. They provide imagery that is unique among the variety of
MRI available today for three main reasons. First, the archive of imagery extends back over 35
years, allowing for longitudinal analyses over a long time span. Second, the imagery is and has
been collected over the entire globe on a regular basis, providing repeat coverage of remote areas
that other satellites do not offer. Third, the imagery is available at no cost.
In the past decade, many changes have occurred both with the Landsat satellites themselves
and with the provision of the imagery. In 1999, Landsat 7 was launched. It provided three years of
high quality data which was complemented by the imagery from Landsat 5, and then sustained a
critical technical problem. The scan-line corrector anomaly (commonly referred to as SLC-off)
reduced the quality and usability of the Landsat 7 data significantly, such that Landsat 5 now
provides the bulk of the imagery. In the next 5 years, the Landsat Data Continuity Mission (LDCM)
is scheduled to be launched, which will replace Landsat 5 in providing primary imagery,
supplemented by Landsat 7 imagery. However, before LDCM is launched, both Landsat 5 and 7
could cease to operate, creating a gap in the provision of imagery. Additionally, the entire catalog
of imagery, including all new acquisitions, became available at no cost at the beginning of 2009,
causing a 50-fold increase in the number of scenes downloaded annually from EROS.
Moderate-resolution imagery, such as Landsat, provides unique spatial information for
many people both within and outside of the United States (U.S.). However, exactly who these users
are, how they use the imagery, and the value and benefits derived from the information are, to a
large extent, unknown. The last comprehensive evaluations were completed over 30 years ago and
1

Policy Analysis and Science Assistance Branch, U.S. Geological Survey, Fort Collins, Colo.

6

DRAFT – DO NOT CITE

attempted to project the conceivable economic benefits of a continued Landsat program (ECON
Inc., 1974; Earthsat, 1974). Much has changed since that time—not only with the capabilities of
remotely sensed data but the applications of the imagery in decisionmaking. More recently, there
have been a small number of studies that have surveyed limited groups of users of MRI (for
example, ASPRS, 2006; EROS, 2007; NSGIC, 2006; Stoney, Fletcher, & Lowe, 2001). While
these surveys have added to the body of knowledge regarding opinions on the attributes of various
sensors, they have not been comprehensive in nature. Our objectives for this study were to 1)
identify and classify the U.S.-based professional users of this imagery; 2) better understand how
and why MRI, and specifically Landsat, is being used; and 3) qualitatively and quantitatively
measure the value and societal benefits of MRI (focusing on Landsat specifically). To reach these
objectives, we conducted a study of U.S.-based professional MRI users from 2008 through 2010 in
two parts: 1) a user identification and 2) a user survey.

User Identification
We defined professional users as those who use MRI in their work. Studying these
professional users is challenging because the extent of the population is unknown and thus, a
representative sample cannot be obtained from the entire population. The membership lists of
various remote sensing and GIS professional organizations could be used, but not every user can be
assumed to be a member of such organizations. The lists of users from imagery providers, such as
EROS (which distributes Landsat), also would not include all the users in the United States. One
reason for this is that data procured by one user is often re-distributed to other users, particularly
when licensing or distribution restrictions do not exist, as is the case with Landsat. Taking a
random sample of the entire population of U.S. residents would not produce usable results, given
that the percentage of people in the U.S. who may be classified as professional users of MRI is
most likely quite low, making this a rare population as well. Because these traditional random
sample selection methods were not adequate or appropriate for this unknown population, we used a
more purposive sampling procedure, explained below.
For this survey, the sample was identified in a two-step process. First, an extensive web
search was performed in the summer of 2008 to identify potential users of MRI in the U.S. The
search was conducted by state and was based on tens of keywords, including remote sensing,
moderate-resolution satellite imagery, and Landsat, along with a variety of application areas and
sectors where use may occur. This search yielded the email addresses of over 20,000 potential users
from across the U.S. Subsequently, snowball sampling was used to confirm the use of MRI and
find additional users. Snowball sampling is a purposive (non-probability) sampling method based
on the existence of social networks. In any population, it can be assumed there are connections
between the members, particularly when that population is specialized in any way, as is the
population of MRI users. Using satellite imagery requires a certain level of technical knowledge
and there are numerous organizations and communities that exist to facilitate the use of satellite
imagery. To take advantage of these social networks, each potential user we identified was
contacted via email, asked to participate in the study, to confirm their use of MRI and to provide
the contact information for up to three other users. These newly identified potential users were
contacted and asked to provide the same information. This process repeated until less than 100 new
users were identified in a “wave.” Snowball sampling is intended to increase the size of a sample
with each wave of new participants, though well-known people within a population tend to be
recommended multiple times.
By the conclusion of the snowball sampling, over 4,000 more email addresses had been
added to the original list of potential users, totaling approximately 25,400 unique email addresses,

7

DRAFT – DO NOT CITE

of which more than 3,500 were undeliverable. Of the almost 22,000 individuals remaining, 5,229
responded to our request and 4,753 agreed to participate in the survey. It is unknown as to whether
the people on the original list were actually users of MRI, so it is not possible to know how many
did not respond to the request. Of those who responded, around 80% were MRI users, 16% were
not users, and 4% were not sure if they were users. The definition of MRI was included in the
snowball sampling email, so the parameters of the imagery were clear. Examples of missions which
collect MRI and products based on MRI were given to further illustrate the definition.

User Survey
We launched the survey in September of 2009 to all the people who agreed to participate.
We developed the survey in conjunction with experts at EROS to ensure that the technical details
were accurate and that the instrument would gather information that would inform the USGS Land
Remote Sensing (LRS) Program’s distribution of imagery and future program requirements. One of
the first decisions we made in this study was to conduct the survey online. An online survey is not
always appropriate for most populations because members cannot be assumed to have access to a
computer, access to the internet, an email account, or the technological skills necessary to complete
a survey online. In this case, the population consisted of imagery users who must have access to a
computer and the internet to use the imagery, who almost certainly have an email account for work
purposes, and who must be at least somewhat technologically adept to use the imagery. Providing
the survey online allowed an opportunity to ask only the questions relevant to each respondent.
Because the sample included both MRI users and non-users, we constructed a survey with
questions tailored to four different types of users:
1) current Landsat users (who had used Landsat in the year previous to the survey),
2) current users of other types of MRI (who had used the imagery in the year previous to the
survey),
3) past users of MRI (who had used the imagery at some point but not in the year previous to
the survey), and
4) users of high- or low-resolution satellite imagery (who had used the imagery at any time).
The set of questions for each of these users was considered a “survey path.” The last two groups
were included because we were interested in understanding why users were using imagery other
than MRI. The answers to certain questions directed respondents to the appropriate survey paths,
reducing the burden on respondents and collecting the most relevant information from each
respondent.

Results
Analysis
We analyzed the data in several different ways, including examining frequency data, chisquare analyses, t-tests, and contingent valuation analyses. Frequencies are reported for many
results and chi-squares are reported for nominal and ordinal data. While many comparisons were
made between various groups in the sample, differences between sectors (e.g., academic institution,
government, private business) were the most prevalent and significant; many of those differences
are reported below. Individual chi squares are reported for each sector where dichotomous
dependent variables exist. Overall chi squares are reported when dependent variables are not
dichotomous. T-tests are reported to compare means computed from scale variables. The contingent
valuation analyses are described in detail in the Willingness to Pay for Landsat Imagery section.

8

DRAFT – DO NOT CITE

Statistical Significance and Interpretation
The sample of professional MRI users in this study is not associated with a known
population of MRI users. Therefore, it is not possible to generalize the results for this sample to the
population as a whole (as is typical when a random sample from a known population is drawn). For
example, the majority of the sample was composed of users of Landsat imagery; however, that does
not mean that the majority of all moderate-resolution imagery professional users based in the U.S.
use Landsat. However, the sample we obtained for this study is robust in terms of its size and the
diversity, such that comparisons among groups within the sample are appropriate. These
comparisons are helpful to illustrate the diversity of users within the sample.
Because of the large sample size, the statistical power of this sample is very high, which
may lead to results that are statistically significant but not meaningfully different (e.g., practically
significant). Because of this, we report statistically significant results at the p < 0.001 level, rather
than the more typical p < 0.05 level found in most behavioral and psychological research.
Additionally, meaningful differences are estimated through measures of association, commonly
called effect sizes. Effect size is a measurement of the amount of impact an independent variable
has on a dependent variable (Murphy & Myors, 1998, p. 12). The effect sizes reported most
2
frequently in this report are phi ( ) and Cramer’s v for chi-square ( ) analyses. Cohen (1988, p. 2527, 79-80) provides the following guidelines for interpreting these effect sizes:
a small effect = 0.1 (i.e., difference in mean height between 15- and 16-year old girls)
a medium effect = 0.3 (i.e., difference in mean height between 14- and 18-year old girls),
and
a large effect = 0.5 (i.e., difference in mean height between 13- and 18-year old girls).
Following Cohen’s recommendations on the interpretation of effect size for behavioral and
psychological studies (1988, p. 25), we consider a statistically significant measure with an effect
size of 0.1 or greater to indicate a meaningful difference for this study. Occasionally, we report
statistically significant results with effect sizes less than 0.1 if they are notable for some other
reason. All statistical results are located in the footnotes.

Diversity of the Sample
Determining the diversity of users within the sample is important in order to begin to
understand what the larger user community might look like. These data cannot be generalized to the
population at large, but the minimum amount of variety in the users and uses within the population
can be gathered from these results. For example, users in the sample applied MRI in over 35
application areas, but users in the population could be applying the imagery in even more
application areas. In other words, while recognizing that this sample may not be representative of
the population as a whole, we believe that the diversity of our sample provides a much needed
baseline of types of users, uses, and valuations which can be expanded with further research.
One of the goals of the study was to reach users outside of the known community and the
results indicate that goal was met. Almost half of the current Landsat users are not members of any
type of remote sensing or GIS organization and, during the snowball sampling, almost 35%
indicated they do not know any other users of MRI. Additionally, 45% of the current Landsat users
did not obtain Landsat imagery from EROS in 2008 or 2009. These results indicate that many of
the users in the sample are not part of the known groups within the population, such as professional
organizations or users who procure imagery from EROS. Aside from reaching beyond the known
user community, the sample was diverse in other ways. Users work in seven different sectors and
use the imagery in over 35 different application areas. They conduct projects at scales from local to

9

DRAFT – DO NOT CITE

global in locations around the world. All of these data indicate that the sample is composed of a
very diverse group of U.S.-based professional users.

User Types
Over 2,500 professionals responded to the survey for a response rate of 53% (n = 2,523).
This response rate is double those typically cited in the literature for online surveys.. The high
response rate may have been due to the initial contact during the snowball sampling, as well as the
high levels of interest and engagement respondents had regarding the topic. Current Landsat users
comprised over half of the sample (fig. 1). Past MRI users comprised 15% of the sample, users of
high- or low-resolution imagery comprised 9% , and current other MRI users represented the
smallest portion of the group with just under 7%. Around 14% had never used any type of satellite
imagery or did not know if they use MRI in their work. The latter two groups were not asked
further questions in this survey because their lack of familiarity with the imagery would have made
it difficult for them to answer the majority of the questions. Around 24% of the respondents had
used Landsat in the past; this group was made up of both current and past MRI users. The survey
was structured to enable comparisons between current Landsat users and current users of other
MRI. However, many other MRI users were unaware of the type of imagery (i.e., Landsat, Terra,
SPOT) they used and other information from the survey (for instance, from open-ended questions)
indicate that many of these “other” MRI users may actually be using Landsat. Because of this and
because the sample was predominantly made up of current Landsat users, the results in the
remainder of the report refer to that group only, except where noted.

Demographics
The average current Landsat user in the sample is male, white, 47 years old, and highly
educated. Three-quarters of the users are male, over 90% are white, over 80% are between 30 and
59 years old, and two-thirds have a masters degree or above. The predominant sector is academic
institutions (33%), followed by private institutions (18%), and then federal (17%), state (16%), and
local government (10%) (fig. 2). Only 4% of the users work for non-profit organizations and less
than 1% work for tribes or nations. Anecdotally, we are aware of many tribes who are using the
imagery and there were many tribal members identified in the initial web search, but very few
responded to the survey. The small number prevented us from comparing respondents in the tribal
sector with those in other sectors.

Use of Landsat Imagery
The first section of the survey established how the current Landsat users in the sample use
the imagery, including types of imagery used, the scales and locations of projects, application areas,
generally how the imagery is used, and the level of use in their work. Each question asked
respondents to consider their use of Landsat in their work over the year previous to the survey.

Types of Imagery
About 40% of respondents indicated they had used only Landsat imagery in the past year.
The remaining 60% of Landsat users indicated they used a mix of imagery, with about half coming
from Landsat, followed by 11% from Terra (ASTER), 8% from SPOT (HRVIR, HRG, HRS), and
3% from Resourcesat (IRS, LISS, AWiFS) (table 1). One percent or less came from ALOS
(AVNIR-2) or CBERS (CCD) on average. Around 6% of the imagery came from other satellites
and about 16% of the imagery was from unknown satellite sources.

10

DRAFT – DO NOT CITE

There are interesting differences among sectors in the imagery used for those using a mix of
MRI (table 1). Respondents in the local government sector used the least amount of Landsat
imagery on average (31%) and used the most imagery from unknown satellite sources (43%).
Academic users using a mix of MRI, on the other hand, obtained 65% of their imagery from
Landsat satellites and only 5% from unknown satellite sources.

Scales and Locations of Projects
Respondents’ projects that relied on Landsat ranged from local to global scales in locations
around the world. Two-thirds of respondents (66%) worked only on projects located in the U.S.,
while 28% have worked on projects in both the U.S. and internationally (fig. 3). Far fewer users
(6%) worked only on projects located internationally. Respondents in the state and local
government sectors were more likely to have worked only in the U.S. than users in other sectors 2 ,
while academic respondents were more likely to have worked in both the U.S. and internationally 3 .
Respondents predominantly worked at the regional scale or lower (fig. 4). Respondents in
4
local government were more likely to have worked at a local scale than other users and those in
state government were more likely to have worked at the state scale 5 .

Application Areas
A list of 37 application areas was developed by examining previous surveys of satellite
imagery users, as well as through consultations with experts at EROS. Respondents were asked to
select their primary application of Landsat from the list, which included an “other” category where
they could write in an application area (table 2). They were then asked to select as many secondary
applications as they wished from the same list. The 37 applications were collapsed into nine larger
categories for the purposes of analysis (table 2). Environmental science and management
applications were the most commonly selected with over 40% of respondents choosing one (fig. 5).
Land use/land cover (17%) was the second most common application, followed by planning and
development (11%), education (8%), and agriculture (8%). Land use/land cover is different than the
rest of the applications since users can be working in environmental science, planning and
development, or any number of other application areas where land use/land cover analyses could be
conducted. Of those who chose land use/land cover as their primary application, the most common
secondary applications were environmental sciences, followed by planning and development
applications, such as urbanization and rural and urban planning and development.
There are clear differences among sectors in these primary applications (fig. 5).
Respondents in the academic sector were more likely to apply Landsat imagery in the area of
6
education , whereas those in the Federal government were more likely to have applied Landsat in
agriculture 7 and environmental sciences 8 . Those in local government were more likely to have used
Landsat for planning and development 9 . Those in the private sector were also more likely to apply

2

Local - 2 = 55.11, Cramer’s v = 0.200; State 2
= 127.88, Cramer’s v = 0.304
4 2
= 30.18, = 0.148
5 2
= 42.26, = 0.175
6 2
= 162.70, = 0.343
7 2
= 28.70, = 0.144
8 2
= 19.98, = 0.120
9 2
= 98.69, = 0.267

2

= 74.98, Cramer’s v = 0.233

3

11

DRAFT – DO NOT CITE

Landsat in planning and development 10 , as well as in commercial applications 11 and
oil/gas/minerals exploration and extraction 12 .
In addition to these current application areas, there may be new and unique uses in the
future. We asked respondents to write in uses they foresee in the next five years. Common
responses included change detection using time series analyses, integration with other imagery or
products, and climate change monitoring and awareness. For example, one respondent believed that
new uses will “mostly come from the power of comparing the long catalog with new observations,
especially associated with urbanization and global warming induced changes.”

How Imagery is Used
To get a sense of how respondents are using MRI in general, we asked them to describe
their overall work with the imagery (fig. 6). The majority of respondents (91%) use the imagery to
answer questions and/or solve problems, process imagery for themselves or others (62%), and
make decisions based on the imagery (57%). Only 19% develop algorithms, 12% provide or sell
imagery or value-added products, and 2% develop commercial software. By sector, those in
13
14
15
academia are more likely to process imagery and develop algorithms whereas those in local
16
and state government are less likely to use the imagery in these ways. Those respondents in the
private sector are more likely to provide or sell imagery 17 and develop commercial software 18 .

Level of Landsat Use in Work
While all the current Landsat users in the sample used Landsat, whether exclusively or in
conjunction with other imagery, the percentage of their work that relied on Landsat over the past
year varied. In order to effectively describe level of use, we categorized respondents as heavy,
medium, or light users. Light users relied on Landsat for 30% or less of their work, medium users
relied on it for 31-70% of their work, and heavy users relied on it for 71% or more of their work.
Overall, almost two-thirds (63%) of respondents were classified as light users, 18% as medium
users and 15% as heavy users (fig. 7). There are some differences between these use levels among
19
sectors , namely in local government. Only 1% of respondents in local government were classified
as heavy users, compared to 10-21% of users in other sectors. In contrast, 81% of local government
users were classified as light users, which is 7-30% higher than in any other sector.
While this categorization is helpful in understanding level of use, it does not indicate
dependence on Landsat. A light user could rely on Landsat for a critical operational use that
accounts for less than 30% of their work, but which would be compromised if Landsat was not
available. We did not explicitly ask about dependence in this survey, though there are some proxies
for dependence that will be discussed later on in the Value of Landsat Imagery section.

10

2

11

2

= 14.89, = 0.104
= 17.43, = 0.112
12 2
= 96.47, = 0.264
13 2
= 17.78, = 0.113
14 2
= 62.33, = 0.212
15
Process imagery - 2 = 22.13, = -0.127; Develop algorithms 16
Process imagery - 2 = 17.00, = -0.111; Develop algorithms 17 2
= 47.80, = 0.186
18 2
= 40.47, = 0.171
19 2
= 104.93, = 0.275

12

2
2

= 19.50, = -0.119
= 32.19, = -0.153

DRAFT – DO NOT CITE

“Local Users”
After examining the results outlined above, it became clear there was a group of users who
were different from the others. We found a group of users characterized by sector, project scale,
and project location, dubbed “local users,” who represent about 25% of the current Landsat users in
the sample. They work for local or state governments applying imagery in local-scale projects
located in the U.S. There are several indications that this group may be a less technical user group
than other users in the sample. They are less likely to process imagery or develop algorithms, more
likely to be a light Landsat user, and less likely to know the satellite source of the imagery they use.
They also appear to be less involved in the professional user community, as they are less likely to
be a member of a remote sensing or GIS-related organization.

Discussion: Use of Landsat Imagery
Overall, Landsat imagery was the primary MRI used by these respondents, but the uses of
the imagery varied greatly among these respondents. In the year previous to the survey,
respondents worked on projects at all different scales in locations around the world. Every one of
the 37 application areas on the list was selected by a minimum of two respondents as their primary
application. The general uses of MRI indicate that there was a mix of technical abilities among the
respondents in the sample, with some respondents processing the imagery or developing algorithms
or software, and others using it in less technical ways to answer questions or make decisions.
There were also respondents using Landsat at all levels in their work, with some using Landsat in
all their work and some using it in very little of their work. Taken together, these results reveal a
diverse sample of users whose responses provide a baseline for exploring the uses in the population
as a whole.
Comparing the results by sector demonstrated that there are significant differences among
the sectors in this sample when it comes to the uses of Landsat imagery. For instance, the existence
of the group of “local users” demonstrates that many respondents in state and local governments
are dissimilar from respondents in other sectors. The results indicate that the different roles and
goals of each sector guide the uses of Landsat imagery by respondents in those sectors.

Changes in Use of Landsat Over Time
There have been many events over the recent history of the Landsat mission that may have
impacted people’s use of the imagery. To track how these events may impact use, respondents were
asked how their use of Landsat changed over the past 10 years and how they envisioned it would
change over the next 5 years.
Around 80% of respondents said their use increased or stayed the same in the past 10 years
and will increase or stay the same in the next 5 years (fig. 8). Of the users who stated their use
increased or would increase, the majority (66-80%) chose both changes in work and cost as
reasons. Respondents cited many other reasons as well, some under the control of data providers
(i.e., availability, accessibility), but most outside of the control of data providers (i.e., fixed
attributes of the sensor like spatial resolution, new applications/uses, more demand for imagery
from clients, improvements in hardware, more training or more knowledgeable staff). Some
respondents based their future increase in use entirely on the successful launch of LDCM and
provision of new usable imagery. By sector, respondents in academia were more likely to cite more
affordable imagery as a reason for increasing use in both the past 20 and the future 21 , whereas those
20

2

21

2

= 40.44, = 0.216
= 24.06, = 0.198

13

DRAFT – DO NOT CITE

in local government were less likely to cite more affordable imagery in both the past 22 and the
future 23 .
Given the recent implementation of the no cost data policy for Landsat imagery and the
anticipated launch of LDCM, increases in the use of the imagery may also occur among
respondents who used Landsat in the past but are not currently using it. Two-thirds of those who
had used Landsat in the past said they foresee using it in the future (fig. 9). When asked what
would make them more likely to use Landsat in the future, the most common reasons were changes
in work and improved spatial resolution. Only 5% said that nothing would make them more likely
to use Landsat.
Less than 13% of respondents said their use had decreased or would decrease (fig. 8). Of the
respondents who said their use decreased over the past 10 years, most cited spatial resolution and
changes in work most often as reasons. However, those who indicated their future use will decrease
cited data quality, temporal resolution, and the attractiveness of other imagery more often. Almost
all respondents who said they will decrease use in the future cite reasons outside of the control of
USGS (96%), including other imagery being more attractive, attributes of the sensor, changes in
work and new Landsat data not being available. Less than 20% cite reasons within the control of
USGS, including availability, accessibility, and licensing/distribution restrictions. Interestingly,
SLC-off on Landsat 7 appears to have had a minimal impact on respondents who said they had
decreased use in the past 10 years – only 69 users cited it as a reason. In response, about two-thirds
of these respondents replaced Landsat 7 imagery with Landsat 5 imagery (fig. 10). Over half
replaced Landsat 7 imagery with other MRI, just under a third used gap-filled or SLC-off Landsat
7, and less than a fifth used some other kind of data as a replacement.

Discussion: Change in Use of Landsat Over Time
Regardless of whether respondents’ use increased or decreased in the past or the future, the
majority of reasons given are outside of the control of USGS, except for more affordable imagery
which was cited only for increasing use. This indicates that most of the current Landsat users in this
sample are satisfied with the provision of imagery by USGS, even if they may want to see changes
to the imagery itself or to the sensors that capture the images. More affordable imagery was cited
by the majority of respondents as a reason for increasing use, except among respondents in the
local government sector. Given that local governments are often faced with restricted budgets, this
is a surprising result. However, it is possible that they have not traditionally paid for any of the data
they use, thus making cost a non-issue. There is also the possibility that they are not aware that
Landsat is available at no cost now (there were several comments in the survey to that effect) and
so do not know that Landsat has become more affordable. Among respondents who decreased use
in the past 10 years, the SLC-off on Landsat 7 was not a major factor. Given that the majority of
these users replaced Landsat 7 with Landsat 5, this indicates that the loss of Landsat 5 before the
launch of LDCM may impact some users.

Impacts of No Cost Data Policy
The entire archive of Landsat imagery became available at no cost at the beginning of 2009.
To determine the impacts of this policy change, we asked respondents about their imagery
acquisitions before and after the policy change (calendar year 2008 and calendar year 2009,
respectively). Respondents did not have to personally download the data, but did have to use it in
22

2

23

2

= 12.35, = -0.119
= 13.08, = -0.146

14

DRAFT – DO NOT CITE

their projects. First we asked where respondents had acquired imagery in 2008 and 2009 and
provided a list of possible sources. EROS was the most common source of the data in both years
for over 45% of respondents, followed by the internet in general (over 20% of users) (fig. 11). A
quarter of the respondents did not know where their imagery came from in both years, indicating
that perhaps they were not personally acquiring the imagery. We were particularly interested in
whether respondents who did not acquire imagery from EROS in 2008 had done so in 2009 in
response to the change in policy. However, less than 10% of the respondents behaved in that
manner and an almost equal amount acquired imagery from EROS in 2008 but not in 2009 (fig.
12).
Though there were few differences in where respondents acquired imagery in 2008 and
2009, there were significant changes in the number of scenes acquired, the percentage of those
scenes acquired from EROS, and the dollar amount spent on scenes between the two years (table
3). A paired samples t-test was conducted on those three variables for which data from both 2008
and 2009 were available for any given respondent. Between 900 and 1,000 respondents had
provided information for both years for each of the three variables. Statistically significant results
were found for each variable. In 2009, the average number of scenes acquired increased by 45%
and the percent of those scenes acquired from EROS increased by 6%. The average amount spent
on Landsat in 2009 was a fifth of what was spent in 2008.

Discussion: Impacts of No Cost Data Policy
Even though there were few changes overall in where respondents acquired their Landsat
imagery before and after the no cost policy went into effect, there were significant changes in how
many scenes were acquired and how much was spent on those scenes in 2008 versus 2009.
Interestingly, the effect sizes (eta-squared in this case) were small for each comparison ( 2 • 0.025).
This means that, for instance, the number of scenes acquired in 2008 accounts for very little of the
variation in the number of scenes acquired in 2009. In this case, the lack of connection between the
variables over the two years indicates that other factors, including the availability of the imagery at
no cost, had a substantial impact on the acquisitions of the respondents in 2009.

Value of Landsat Imagery
In economic terms, the value of information is equal to what individuals would pay for that
information (Macauley, 2006). The value depends on the uncertainty of the situation in which the
information will be used, the importance of the outcome of the situation, the cost of using the
information, and the cost of an appropriate substitute. Macauley (2005, 2006) notes that there are
several ways the economic value of information has been examined, including output or
productivity measures, hedonic price studies, contingent valuation studies, and measurement of
societal benefits. However, societal benefits can be difficult to measure economically, especially
when the realized value is in relation to a nebulous, but important, concept like quality of life.
Additionally, the comprehensive value of Landsat may always be elusive, given the widespread use
of the imagery in applications like Google Earth and the difficulty in finding all direct and indirect
users of the imagery. All of these factors emphasize the importance of measuring the value of
information provided by Landsat imagery in multiple ways.
We used four approaches to estimate the value of Landsat to this sample of professional
Landsat users. First, we explored the importance of Landsat imagery to respondents, as well as their
satisfaction with the imagery. Second, we asked about the environmental and societal benefits,
including impacts on decision-making, from projects that used Landsat. Third, we asked what
respondents would do if Landsat imagery was no longer available and how it would impact their

15

DRAFT – DO NOT CITE

work. Lastly, we utilized a method called contingent valuation to determine respondents’
willingness to pay for imagery equivalent to Landsat in the event that there is a gap in imagery
provision.

Importance and Satisfaction
Exploring the importance of Landsat imagery to users is one way to approach value. More
than 80% of the respondents said the imagery is somewhat or very important to their work (fig. 13).
Once again, there are differences by sector. Respondents in academia are more likely to think the
24
25
26
imagery is very important to their work while those in the state and local government sectors
are less likely to say Landsat is very important.
We also asked respondents to rate how important certain attributes of MRI are in
determining which type of MRI to use and how satisfied they are with those same attributes as they
occur in Landsat. This is a common approach in marketing research to assess how well a product is
meeting the needs of customers (Martilla & James, 1977). From this data, we created an
importance-performance framework that maps satisfaction on the X axis by importance on the Yaxis (fig. 14). It allows us to look at where things are going well and where room for improvement
exists. All of the Landsat attributes we asked about fall in the “keep up the good work” quadrant; in
other words, on average users think all of the attributes measured are important and they are
satisfied with the provision of those attributes. The highest ratings are for availability, accessibility
and cost, which indicate that users are satisfied with how the imagery is being provided. There were
no significant differences in these ratings by sector, application area, or any other variables, except
for global coverage. Global coverage is different from the rest of the attributes because it was rated
as having only average or neutral importance. We believe this is driven by the large portion of the
sample that is doing work only in the U.S. and do not need international coverage. This is
supported by the fact that global coverage is more important for people working internationally:
only 23% of people working in just the U.S. think it is important, whereas 63% of people working
internationally think it is important 27 .

Benefits of Landsat
When discussed in most venues, the benefits of Landsat are typically assumed or only
anecdotally documented. In an effort to gather more systematic information about the benefits of
Landsat, we asked a series of open-ended questions where respondents could write in their
responses. Open-ended questions were chosen because no comprehensive list of benefits has been
developed and we wanted to give respondents the opportunity to provide their own ideas about
benefits. The responses were examined for repeating themes and then categorized into those
themes. When asked how projects using Landsat had affected decisionmaking, the most common
responses were through more informed decisionmaking, impacts on policies or regulations, and
better planning and management. One respondent noted, “We have been able to come up with
evidence to change small town policy and challenge politics. Good science is hard to beat.”
A second question inquired about the environmental and/or societal benefits that have come
about as a result of projects which used Landsat. Assessing impacts and change over time,
habitat/land conservation, improving the environment/reducing impacts, increasing human
safety/reducing risk to humans, and better decision-making were among the most frequently cited
24

2

25

2

26
27

= 55.71, Cramer’s v = 0.207
= 27.06, Cramer’s v = 0.144
2
= 42.35, Cramer’s v = 0.181
2
= 276.58, Cramer’s v = 0.427

16

DRAFT – DO NOT CITE

benefits. One respondent said “Perhaps Landsat[‘s] greatest utility is the long time frame that can
be accessed.” Another stated, “The greatest benefit I have perceived is that providing stakeholders
with spatial data gives them a sense of empowerment to make decisions. The unexpected benefit is
that map data act[s] to build consensus among groups with competing interests. Seeing actual data
tends to dissolve apparent symbolic differences and helps different groups work together on issues
of environmental management.”

If Landsat Was No Longer Available…
One of the ways to examine the value of a good is to explore the impacts that would occur if
it ceased to exist. We asked users what would happen to their work if both new and archived
Landsat was no longer available. We assumed users could:
• discontinue some or all of their work;
• continue their work without substituting other imagery or information; or
• use other imagery or information as a substitute in their work.
Around half of the users would discontinue at least some of their work and would continue
at least some of their work without substituting other imagery or information (table 4). Just over
75% of the users would substitute either other imagery, other data sets, or field work for at least
some of their work.
Of those who would discontinue some of their work (n = 693), 45% would discontinue a
small percentage of their work (30% or less), 32% would discontinue a medium percentage (3170%), and 23% would discontinue a high percentage (71% or more). In fact, 11% would
discontinue over 90% of their work if Landsat was no longer available, which indicates a strong
dependence on the imagery. There were some differences by sector on this variable. Academic
users were more likely to discontinue a medium to high percentage of their work 28 than users in
other sectors.
Of those who would substitute, almost 90% would use other imagery as a substitute, about
two-thirds would use other data sets and slightly fewer would use fieldwork (table 5). Given that
fieldwork tends to be more expensive and time consuming than using imagery or other data, this
seems to indicate that fieldwork might be the only viable substitute to provide certain types of data.
This may be because other similar appropriate imagery does not exist or is not affordable. If users
indicated they would use substitute imagery, they were asked what imagery they would prefer
regardless of budget constraints. Then they were asked what imagery they would most likely
acquire given budget constraints. More than half would choose the same imagery regardless of
budget constraints – the most common choice was Terra (ASTER), followed by SPOT, and
Resourcesat (fig. 15). However, 40% would choose different imagery. SPOT was the preferred
imagery for over half of these users, but Terra was the imagery that over half of them would be
most likely to acquire (fig. 16).
Another way to explore value is to examine what would happen to costs and
revenues/funding if Landsat were no longer available. Increases in costs could occur, for instance,
if users have to pay for other imagery, data, or field work to replace the information provided by
Landsat imagery. Revenues could possibly decrease because a product based on Landsat can no
longer be produced or the product is created from a more expensive type of data. Typically, these
sorts of budgetary questions can only be answered by certain individuals in an organization who
have access to that information. We knew that not everyone in this sample would be able to
respond to these questions and therefore, we only asked for information regarding the projects in
28

2

= 49.98, Cramer’s v = 0.216

17

DRAFT – DO NOT CITE

which the respondents were involved. Respondents also had the option to indicate they did not
know.
When asked to estimate the percentage increase in costs for their Landsat-related projects,
almost half of the respondents felt their costs would increase by at least 1%, 41% didn’t know if
their costs would increase, and 11% felt their costs would not increase. Of those who believed their
costs would increase (n = 668), the average total percent increase in costs was 30%, which
translated to about $27,000 on average among those who were able to provide the current costs of
all their projects which rely on Landsat imagery (n = 519). When asked about certain types of cost
increases, over half of the respondents said it is somewhat or very likely that total, processing, and
administration/overhead costs would increase and that more time would be spent on projects.
However, most believed it is unlikely that they would purchase additional equipment/software or
hire more staff.
Regarding changes in revenues/funding, 43% of users did not know what impact the loss of
Landsat would have. A third felt there would be no impact and a fifth thought their
revenues/funding would increase.

Willingness to Pay for Landsat Imagery
Economic benefits, whether of a market or non-market good, are measured by the
maximum amount the users would pay for another unit of it. For market goods, price measures the
willingness to pay (WTP) for one more unit (e.g., a pound of apples, a gallon of gas). Estimating
the user benefits for non-market and publicly provided goods can be challenging since there is
either no price (Landsat imagery is currently available at no cost) or only an administrative price
(as was the case when Landsat imagery was provided for a fee). The administrative price often does
not reflect market forces or an equilibrium price and quantity, and sometimes will substantially
under or overstate what a user would pay. A single administrative price will only reveal a single
point on the user’s demand for the good. Using WTP for both market and non-market goods
ensures commensurability between dollar benefit estimates of these two types of goods.
WTP is the standard measure of benefits in benefit cost analysis (Sassone and Shaefer,
1978) and economists use a variety of techniques to estimate the WTP for non-market goods.
Champ and others (2003) provide an accessible review of each of the commonly used methods.
When there is no price, or there is little data available on what users will pay, a stated preference or
intended behavior technique known as the Contingent Valuation Method (CVM) is commonly
used. This method uses a simulated or hypothetical market to determine the maximum amount a
user would pay for another unit of the good rather than use a more expensive substitute or do
completely without. The method is recommended for use by federal agencies performing benefit
cost analysis (Office of Management and Budget, 1992; U.S. Environmental Protection Agency,
2000; U.S. Water Resources Council, 1983). As suggested by the National Oceanic and
Atmospheric Administration panel on contingent valuation (Arrow and others, 1993), we asked a
dichotomous choice format question. In this case, the user must only decide whether the Landsat
scene is worth more than the cost specified in the question. The specific question asked was:
“If Landsat 5 and 7 became inoperable before the next Landsat satellite is operational
(scheduled to launch in 2012), you may have to obtain imagery elsewhere during the
interim. Assume that you are restricted to your current project or agency budget level and
that the money to pay this cost would have to come out of your existing budget. If such a
break in continuity did occur and you had to pay for imagery that was equivalent to the
Landsat standard product now available, would you pay $XXX for one scene covering the
area equivalent to a Landsat scene?”

18

DRAFT – DO NOT CITE

The “$XXX” was filled in with one of 21 different dollar amounts. The dollar amounts
ranged from a low of $5 to a high of $5000. To measure an individual’s maximum WTP based on
Yes/No responses to a single dollar amount the goal was to have a dollar amount low enough that
nearly all Landsat users would answer Yes, and a dollar amount high enough that nearly all Landsat
users would answer No. The remaining dollar amounts were $10, 20, 30, 40, 50, 60, 70, 80, 100,
150, 200, 250, 300, 400, 500, 700, 1200, 1500, and 2500. The response to this question provides
the data necessary to calculate a single-bounded WTP. The response to this question also allows the
estimation and development of a demand function for the nonmarket good – as opposed to the
single point on the demand function from administrative pricing. This WTP question and approach
includes an explicit budget constraint and a reminder that the funds to pay the higher cost would
have to come out of this fixed budget. This follows the recommendation of the NOAA panel
(Arrow and others, 1993) that budget reminders are to be included in WTP questions. Recognition
of budget constraints is important to be consistent with consumer behavior and demand theory.
The precision of the estimate of WTP can be increased by asking respondents a second
dichotomous choice question, thus collecting the data necessary for a double-bounded estimate of
WTP. With the first WTP response, we know that the respondent’s WTP is higher or lower than the
dollar amount asked to pay. If those who answer Yes to the first question are asked if they would
pay a higher amount, we can obtain additional information about their WTP: (a) if they state “Yes”
to the second, higher dollar amount, we know their WTP is higher than the first bid amount and in
fact is even higher than the increased bid amount (the second dollar amount is twice the first in this
case); (b) if they state they would not pay this second higher dollar amount we know their WTP is
bounded between the first and second dollar amount, thus bracketing their maximum WTP between
the first and second dollar amount. The same pattern holds for those answering “No” to the first
dollar amount. If a respondent answers “Yes” to the second dollar amount (half the first dollar
amount in this case), we can potentially bracket their WTP between the first and second dollar
amount. This second question can substantially improve the efficiency of estimating the WTP but
risks offending the respondent with a moving price target.
Finally, after the first and second WTP questions, respondents were asked how certain they
are that they would (or would not) pay that price on a scale of 0% to 100%. This provides further
information about the actual WTP of each respondent and allows for examination of the quality of
each respondent’s answers. For example, if a respondent says “Yes” to the first WTP amount and
“Yes” to the second higher amount, economic logic suggests they should be less certain (or the
same) about paying the higher amount.
Single-Bounded WTP Analysis

The demand curve derived from the responses to the initial question is shown in figure 17.
This demand curve can be used to determine the median and mean WTP for this sample of users.
The median is the amount where a typical user crosses over between purchasing and not purchasing
– in this case, it is approximately $218 per scene. It is not the value of the scene to the user but the
value where a majority (>50%) of the sampled users would purchase a scene equivalent to a
Landsat scene. The mean value of a scene involves integrating the area under the demand curve,
which is a cumulative distribution function in this case. For the curve shown in figure 17, the
average is unbounded. The value of Landsat to some users is substantial and this pulls the average
well above the median. The standard practice when faced with an unbounded average is to choose
one of the high bids in the sample to use for the upper bound in the integration. We chose the
second highest bid ($2,500) and calculated a mean WTP of $760 per scene. Choosing the highest
bid would increase the mean and choosing the third highest bid would decrease it. A practical
implication of the unbounded mean WTP is that we need to increase the number of bid amounts at
19

DRAFT – DO NOT CITE

the middle to higher end of the range and decrease the number of bids at the lower end in future
questions. More information about respondent’s WTP at these high bids would allow better
estimation of the complete demand function.
When WTP is broken out among sectors, significant differences can be seen between the
means of the groups (fig. 18). Federal government users are willing to pay the most at over $1,000,
followed by academic users at over $800, and private users at around $700. Again, because the
sample is not random, neither the median nor the mean WTP for this sample or any of the groups
within the sample should be generalized to the population as a whole or any segments of the
population.
Double-Bounded WTP Analysis

In concept, the double-bounded analysis is intuitively appealing as statistical theory and
past studies have shown that asking the second WTP question does reduce the variance of WTP
estimates and gives more precision. However, we found that the respondents’ behavioral response
to the second bid amount is somewhat different than the response to the first bid amount. For
example, one respondent said “No” to $150 with a certainty of 40%, but then said “No” to $75 with
a certainty of 60%. In practice, it appears that respondents do not like the follow-up question with
the different bid amount. The information gathered from the second question is not as good as the
information from the first question. For both the first and second questions, the probability of
saying “Yes” goes down as the bid amount increases, but at somewhat different rates in response to
each question. There are several reasons this may be the case, including strategic behavior on the
part of respondents, uncooperative behavior, or changed preferences. At this time, we hypothesize
that the behavioral shift between first and second responses may be due to the large magnitude of
the increase between the first and second bids. Especially at high first dollar amounts ($1500,
$2500, and $5000), the very large absolute magnitude of the increase in the second dollar amount
($3000, $5000 and $10,000) may have reduced the credibility of the second WTP question to the
respondent. It is possible that the response to this second question was more of a signal about the
question itself rather than a second “snapshot” of their WTP.
As shown in figure 19, the demand curve of the double-bounded analysis does behave as
expected when compared to the curve of the single-bounded analysis. The double-bounded curve
shifts to the left and has much tighter confidence intervals than the single-bounded curve.
There are no significant differences found between sectors in the double-bounded analysis.
The reasonable differences in WTP across groups of respondents disappear with the information
from the second question. The standard errors on the sector variables were larger than the
parameter estimates so the variables can reasonably be removed from the double-bounded model.
Discussion: Single-Bounded Vs. Double-Bounded WTP Analyses

The double-bounded analysis improves efficiency but the second question does not have
as much value as the first question due to the behavioral shift between questions. This behavior was
not expected among Landsat users, since they have experience with the imagery and know the good
being valued quite well in most cases. Given the results, we chose to report the standard singlebounded dichotomous choice CVM results because they are the best in terms of the statistical
significance of independent variables, goodness of fit (the percent of correct predictions is about
70%), and the method is a well-accepted industry standard.
Several things might be tried for a future WTP for Landsat imagery question to reduce the
behavioral difference between the first and second questions. The magnitude of the change between
the first and second bid could be reduced from twice and half to 1.25 and 0.75. The number of

20

DRAFT – DO NOT CITE

initial high bid amounts could be increased since these higher amounts are essential to accurate and
precise estimates of mean WTP which involves integration across the entire demand function.
Other statistical modeling techniques such as ordered logit models which allow for one category for
each of the four combinations of responses (Yes-Yes, Yes-No, No-Yes, and No-No) could also be
tried.

Conclusion
The results of the survey revealed that respondents from multiple sectors use Landsat in
many different ways, as demonstrated by breadth of project locations and scales, as well as
application areas. The current level of use will likely continue and more likely to increase among
these respondents, particularly as it becomes better known that the imagery is available at no cost
and as new uses are identified. The changes in acquisition patterns, including the increase in the
number of scenes acquired and the decreasing amount spent after the imagery became available at
no cost, also point toward increases in future use.
The value of Landsat imagery is high overall for these respondents, and those in certain
sectors, such as academia, find it even more valuable than respondents in other sectors. In general,
Landsat imagery is important to respondents for their work and they are very satisfied with the
attributes provided by Landsat. They find it beneficial for improving decision making and
preventing harm to the environment and humans, among many other benefits. These benefits will
likely increase as the no cost policy change becomes widely known, LDCM is launched, and
emerging issues facing the nation, such as climate change, become more pronounced and require
increasing amounts of longitudinal, reliable data. The value of Landsat imagery to these
respondents is also demonstrated by the amount of work that would be discontinued or require a
substitute source of data. Finally, the value is demonstrated by respondents’ willingness to pay for
the good. On average, respondents are willing to pay $760 per scene, which is greater than the
previous administratively set price.
Throughout the analyses, differences between sectors occurred on almost all variables. The
identification of the “local users” group help to explain some of these differences, but respondents
in the academic and federal government sectors also tend to differ from the rest of the respondents.
They tend to be more technical users, use the imagery more heavily, and are willing to pay more for
replacement imagery. These differences between sectors merit further exploration in future studies
to see whether they exist within the other segments of the population as well.
The survey collected information from users who are both part of and apart from the known
user community. The diversity of the sample delivered results that provide a baseline of knowledge
about the users, uses, and value of Landsat imagery. While the results supply a wealth of
information on their own, they can also be built upon through further research to generate a more
complete picture of the population of Landsat users as a whole. Surveying other segments of the
Landsat user population, such as those who procure imagery from EROS and international users,
and conducting case studies on the value of Landsat imagery in specific applications would provide
additional useful information.

21

DRAFT – DO NOT CITE

References Cited
Arrow, K., Solow, R., Portney, P.R., Leamer, E.E., Radner, R., and Schuman, H., 1993, Report of
the NOAA panel on contingent valuation: Wash., D.C., National Oceanic and Atmospheric
Administration, U.S. Department of Commerce.
ASPRS, 2006, Survey on the future of land imaging: Bethesda, Md., American Society for
Photogrammetry and Remote Sensing.
Berg, S., 1988, Snowball sampling in Kotz, S. and Johnson, N. L. (eds.), Encyclopedia of Statistical
Sciences, v. 8, p. 528–532.
Browne, K., 2005, Snowball sampling—Using social networks to research non-heterosexual
women: International Journal of Social Research Methodology, v. 8, no. 1, p. 47–60.
Champ, P.A., Boyle, K.J., and Brown, T.C. (Eds.), 2003, A primer on nonmarket valuation (The
economics of non-market goods and resources series, Bateman, I.J. (Ed.)): Norwell, Mass.,
Kluwer Academic Publishers, 577 p.
nd
Cohen, J., 1988, Statistical power and analysis for the behavioral sciences (2 ed), Hillsdale, N.J.,
Lawrence Erlbaum Associates, Inc.
Dillman, D.A., 2007, Mail and internet surveys—The tailored design method, 2nd edition: Hoboken,
N.J., John Wiley & Sons, Inc., 523 p.
EarthSat Corp., 1974, Executive summary—Earthsat study on the value of ERTS information.
ECON Inc., 1974, The economic value of remote sensing of earth resources from space—An ERTS
overview and the value of continuity of service: Princeton, N.J., ECON Inc.
EROS (Earth Resources Observation Science Center), 2007, Customer satisfaction survey—
Landsat data: Sioux Falls, S.Dak., Earth Resources Observation Science Center, 78 p.
FLIIWG (Future of Land Imaging Interagency Working Group), 2007, A plan for a U.S. national
land imaging program: Wash., D.C., Office of Science and Technology Policy and National
Science and Technology Council, 120 p.
Macauley, M.K., 2005, The value of information—A background paper on measuring the
contribution of space-derived earth science data to resource management (Discussion paper 0526): Wash, D.C., Resources for the Future, 27 p.
Macauley, M.K., 2006, The value of information—Measuring the contribution of space-derived
earth science data to resource management: Space Policy, v. 22, p. 274-282.
Martilla, J.A., and James, J.C., Importance-performance analysis. Journal of Marketing, v 2, no 1,
p. 77-79.
Murphy, K.R., and Myors, B., 1998, Statistical power analysis—A simple and general model for
traditional and modern hypothesis tests: Mahwah, N.J., Lawrence Erlbaum Associates, Inc.
Office of Management and Budget, 1992, Guidelines and discount rates for benefit-cost analysis of
federal programs, Circular A-94, Revised Transmittal Memo #64: Wash., D.C., Executive Office
of the President.
NSGIC (National States Geographic Information Council), 2006, Results of a national survey—
Imagery for the Nation Proposal: Bel Air, Md., National States Geographic Information Council,
40 p.
Sassone, P., and Schaffer, W., 1978, Cost benefit analysis—A handbook: New York, NY,
Academic Press.
Stoney, W., Fletcher, A., and Lowe, A., 2001, Data only report of Landsat user survey (draft): Falls
Church, Va., Mitretek Systems, 85 p.
U.S. Environmental Protection Agency, 2000, Guidelines for preparing economic analyses, EPA
240-R-00-003: Wash., D.C., U.S. Environmental Protection Agency.

22

DRAFT – DO NOT CITE

U.S. Water Resources Council, 1983, Economic and environmental principles and guidelines for
water and related land resources implementation studies: Wash, D.C., U.S. Government Printing
Office.

23

DRAFT – DO NOT CITE

Figure 1. Types of imagery users.

60%

Current Landsat user

1392

Current other MRI user
Past MRI user

50%

High/low user

Percent of respondents

Not a user
Don’t know

40%

Past Landsat user

30%
445
20%
383
10%

229

241

167

211

0%
Type of user

24

DRAFT – DO NOT CITE

Figure 2. Sectors of respondents who are currently using Landsat.
35%

Academic institution
Federal gov

30%

State gov

Percent of respondents

Local gov
Private business

25%

Non‐profit org
Tribe or nation

20%

Other

15%

10%

5%

0%
Current Landsat user

25

DRAFT – DO NOT CITE

Figure 3. Locations of projects using Landsat imagery in year previous to survey.

100%

All sectors
Academic institution

90%

Federal government
State government

80%

Local government

Percent of respondents

Private business

70%

Non‐profit organization

60%
50%
40%
30%
20%
10%
0%
United States only

International only

Both US and international

26

DRAFT – DO NOT CITE

Figure 4. Scales of projects using Landsat imagery in year previous to survey.

100%

All sectors
Academic institution

90%

Federal government
State government

80%
Percent of respondents

Local government

70%

Private business
Non‐profit organization

60%
50%
40%
30%
20%
10%
0%

27

DRAFT – DO NOT CITE

Figure 5. Applications of Landsat imagery in year previous to survey.

70%

All sectors
Academic institution

60%

Federal government
State government
Local government

50%
Percent of respondents

Private business
Non‐profit organization

40%
30%
20%
10%
0%

28

DRAFT – DO NOT CITE

Figure 6. General use of moderate-resolution imagery among respondents who are currently using
Landsat.

100%

All sectors
Academic institution

Percent of respondents

90%

Federal government

80%

State government

70%

Private business

Local government
Non‐profit organization

60%
50%
40%
30%
20%
10%
0%
Use imagery
to answer
questions

Process
imagery

Make
decisions
based on
imagery

Develop Provide or sell Develop
algorithms
imagery
commercial
software

29

DRAFT – DO NOT CITE

Figure 7. Level of use of Landsat imagery in work in year previous to survey.

90%

All sectors
Academic institution

80%

Federal government
State government

Percent of respondents
Percent of respondents

70%

Local government
Private business
Non‐profit organization

60%
50%
40%
30%
20%
10%
0%
Light use

Medium use

Heavy use

30

DRAFT – DO NOT CITE

Figure 8. Change in use of Landsat imagery in past 10 years and next 5 years.

60%

Past 10 years
Next 5 years

Percent of respondents

50%

40%

30%

20%

10%

0%
Increase use

Use stays the
same

Decrease use

31

Cannot
characterize use

DRAFT – DO NOT CITE

Figure 9. Potential use of Landsat imagery among past Landsat users.

60%
Past Landsat user

Percent of respondents
Percent of respondents

50%

40%

30%

20%

10%

0%
None

Landsat only

Landsat & other MRI

32

Other MRI only

DRAFT – DO NOT CITE

Figure 10. Reactions to SLC-off on Landsat 7.

80%

All sectors
Academic institution

70%

Federal government
State government
Private business

Percent of respondents

60%
50%
40%
30%
20%
10%
0%
Replaced L7
imagery with L5
imagery

Replaced L7
imagery with
other MRI

Use gap‐filled L7
imagery

33

Replaced L7
imagery with
other kinds of
data

Still use some
SLC‐off L7
imagery

DRAFT – DO NOT CITE

Figure 11. Sources of Landsat imagery acquisitions before and after the imagery became available
at no cost (2008 versus 2009).

50%

2008

45%

2009

Percent of respondents

40%
35%
30%
25%
20%
15%
10%
5%
0%

*Other sources include universities, federal government agencies (besides USGS), state governments, commercial entities, 
and international entities.  This category also includes respondents who did not obtain any imagery during that year.

34

DRAFT – DO NOT CITE

Figure 12. Patterns of acquisitions of Landsat imagery from EROS before and after it became
available at no cost (2008 versus 2009).

50%
45%

Percent of respondents

40%
35%
30%
25%
20%
15%
10%
5%
0%
2008 & 2009

2008 only

2009 only

35

Neither 2008 or 2009

DRAFT – DO NOT CITE

Figure 13. Importance of Landsat imagery to the work of respondents.

60%

All sectors
Academic institution
Federal government
State government

50%

Local government

Percent of respondents

Private business
Non‐profit organization

40%

30%

20%

10%

0%
Very
unimportant

Somewhat
unimportant

Neither
important nor
unimportant

Somewhat
important

36

Very important

DRAFT – DO NOT CITE

Very 
Important

Figure 14. Importance of and satisfaction with certain attributes of Landsat.

5
“Concentrate Here”

Importance

4
“Keep Up the Good Work”

3

2
Very
Unimportant

“Low Priority”

“Too Much Effort Here”

1
1
Very 
Dissatisfied

2

3

4

Satisfaction

Accessibility
Area/footprint of individual scene
Cost
Delivery time
Global coverage
Spatial resolution
Temporal resolution/frequency of coverage

5
Very 
Satisfied

Archive/continuity
Availability
Data quality assessments
Ease of use
Licensing/distribution restrictions
Spectral resolution

37

DRAFT – DO NOT CITE

Figure 15. Preferred imagery versus imagery most likely to acquire within budget constraints
among respondents who would substitute other imagery for Landsat if it was no longer available.

50%

Preferred imagery
(no budget
constraints)

45%
40%

Imagery most
likely acquired
(within budget
constraints)

Percent of respondents

35%
30%
25%
20%
15%
10%
5%
0%

38

DRAFT – DO NOT CITE

Figure 16. Preferred imagery versus imagery most likely to acquire within budget constraints
among respondents who would choose a different imagery based on budget constraints.

60%

SPOT
Terra
Resourcesat

50%

High res

Percent of respondents
Percent of respondents

ALOS
Other MRI
CBERS

40%

Other
Don't know
Low res

30%

Free/cheapest
None

20%

10%

0%
Preferred imagery

Imagery most likely to acquire

39

DRAFT – DO NOT CITE

Figure 17. Single-bounded demand curve of willingness to pay for imagery to replace Landsat.

Per Scene Willingness-to-Pay Amounts

$5,000
Single-Bounded
Curve

$4,500
$4,000
$3,500
$3,000
$2,500
$2,000
$1,500
$1,000
$500
$0
0%

20%

40%
60%
Percent of "Yes" Answers

40

80%

100%

DRAFT – DO NOT CITE

Figure 18. Single-bounded logit model results for willingness to pay for imagery to replace
Landsat.

$1,200
Median

Willingness-to-Pay

$1,000
Mean
$800
$600
$400
$200
$0

41

DRAFT – DO NOT CITE

Figure 19. Single- and double-bounded demand curves of willingness to pay for imagery to
replace Landsat.

Per Scene Willingness-to-Pay Amounts

$5,000
Single-Bounded
Curve
Lower CI

$4,500
$4,000

Upper CI

$3,500

Double-Bounded
Curve
Lower CI

$3,000

Upper CI

$2,500
$2,000
$1,500
$1,000
$500
$0
0%

20%

40%
60%
Percent of "Yes" Answers

42

80%

100%

DRAFT – DO NOT CITE

Table 1. Mean percent of different types of imagery used in the year previous to the survey among
respondents who used a mix of moderate-resolution imagery.
Imagery

All sectors

Academic
institution

Federal gov

State gov

Local gov

Private

Non-profit
org

Landsat

54%

65%

57%

49%

31%

48%

57%

Terra

11%

15%

12%

7%

5%

10%

11%

SPOT

8%

7%

8%

9%

8%

7%

9%

Resourcesat

3%

2%

7%

2%

2%

3%

4%

ALOS

1%

<1%

1%

1%

<1%

2%

2%

CBERS

<1%

<1%

1%

<1%

<1%

1%

1%

Other

6%

5%

6%

5%

10%

8%

3%

Unknown

16%

5%

8%

26%

43%

21%

13%

Total

100%

100%

100%

100%

100%

100%

100%

43

DRAFT – DO NOT CITE

Table 2. Applications of Landsat imagery.
Collapsed applications for analysis

Individual applications

Agriculture

Agricultural forecasting
Agricultural management/production/conservation

Environmental sciences and management

Biodiversity conservation
Climate science/change
Coastal science/monitoring/management
Ecological/ecosystem science/management
Fish and wildlife science/management
Fire science/management
Forest science/management
Geology/glaciology
Range/grassland science/management

Land use/land cover
Planning and development

Recreation science/management
Water resources (e.g., watershed management, water rights,
hydrology)
Land use/land cover
Assessments and taxation
Engineering/construction/surveying
Rural planning and development
Urban planning and development
Urbanization

Commercial

Cultural resource management (e.g., archaeology, anthropology)
Real estate/property management
Software development
Telecommunications
Transportation
Utilities

Education

Education: K-12
Education: university/college
Technical training (e.g., workshops, short courses)

Human needs

Emergency/disaster management
Hazard insurance (e.g., crop, flood, fire)
Humanitarian aid
Public health

Legal/security

Defense/national security
Environmental regulation
Law enforcement

Oil/gas/minerals

Oil and gas/mineral exploration/extraction

44

DRAFT – DO NOT CITE

Table 3. Acquisitions of Landsat imagery before and after imagery became available at no costs
(2008 versus 2009).
Acquisition variable

2008 mean

2009 mean

t

p

95

138

-3.95

<0.001

41%

47%

-4.93

<0.001

$5,117

$1,163

5.28

<0.001

Number of scenes acquired
Percent of scenes acquired from EROS
Dollars spent on imagery

Table 4. Percentages of respondents who would take each of three actions if Landsat imagery
was no longer available.
Action taken if Landsat was no longer available

Would take
action

Would not take
action

Don’t
know

Substitute other imagery or information in work

76%

4%

20%

Discontinue work

51%

28%

21%

Continue work without substituting other imagery or
information

46%

30%

24%

Table 5. Percentages of respondents who would use each of three types of information as
substitutes for Landsat imagery if it was no longer available.
Type of substitute information

Would use

Would not use

Don’t know

Other imagery

89%

1%

10%

Other data sets

69%

15%

16%

Fieldwork

63%

25%

12%

45