Basic Concepts and Terminology

steamgloomyElectronics - Devices

Nov 15, 2013 (3 years and 9 months ago)

87 views

Remote Sensing Systems and

Digital Image Data


Basic Concepts and Terminology

o
Digital image structure


o
Resolutions

o
Spatial

o
Spectral

o
Radiometric

o
Temporal


o
Types of sensors


o
Platforms


o
Proximal (non
-
imaging) remote sensing



Several
images of
the same ground area
for
different regions of the EM spectrum (e.g
.,
blue, green, red, near
-
infrared
).

Each
p
ixel
location in an array can be identified using
row

(i)
and
column

(j) designation.

‘Multi
-
spectral’ image data contains multiple
bands

(k)

of
individual
raster images associated with different regions of the EM spectrum.


Digital remote sensing data is usually stored as a matrix (or raster) of numbers
associated with pixels that form a digital image. A
‘pixel’


a 2
-
dimensional
picture element that is the smallest non
-
divisible element (minimum unit of
information) of a digital image.

Blue


Green


Red

NIR

Blue


Green

Red

NIR

Each pixel within the image data has a ‘brightness value’ (BV)
associated with
each spectral band, which are geographically registered to one another.

Blue


Green


Red

NIR

Blue


Green

Red

NIR

Brightness Value (BVs
)
(also commonly called digital numbers [DNs] or
digital counts/values)
represent the optical properties (reflected or emitted
radiation) of the surface area sampled with a sensor’s instantaneous field of
view (IFOV) converted to a digital value.

Pixels and their associated BVs form the
basis for the interpretation, analysis, and
acquisition of information about surface
features (or objects) for most agricultural
and natural resource applications.

BVs of a single spectral band associated with an array of pixels from
a spatial subset of the Landsat Thematic Mapper (TM) image.

Resolutions of a Digital Image

Resolution



a measure of the ability of a sensor system to
discriminate signals that are spatially near or spectrally similar.


Four type of resolution
associated with digital image data:


1. Spatial resolution


2. Spectral resolution


3. Radiometric resolution


4. Temporal resolution

Airborne multi
-
spectral digital image over forest canopy.

Resolutions of a Digital Image

1.
Spatial Resolution



measure of the smallest angular
separation between two objects that can be distinguished
(
smallest object that can be detected on an image
).


o
Determined by the
instantaneous field of view



(IFOV)



the angular section of the Earth’s


surface observed by the sensor
.


o
Resolution depends on the sensor’s:

1.
Height (altitude)

2.
Speed of data collection

3.
Number of detectors



Today’s satellite
sensors have spatial resolutions
ranging from several kilometers to less than one meter
.
Airborne sensors have higher spatial resolution on the
order of a few meters or less.

MODIS 250
-
meter

multi
-
spectral image

Landsat 30
-
meter

multi
-
spectral image

IKONOS 4
-
meter

multi
-
spectral image

Geographic scope and target of
interest for an application will
determine appropriate remote
sensing image data.

The smaller the ground
area represented by a
pixel, the higher the
spatial resolution of the
image.


Selection of appropriate spatial resolution
is closely related to the suitable scale of
the particular problem under study.

Spatial resolution affects the “purity” of
the spectral signal of a pixel
.



The
smaller the size of the pixel
,
the
higher probability that the pixel’s
signal will be representative of a
single target
or land cover type.


The
larger the pixel size
, the
higher
probability that the pixel’s spectral
signal will be average signal of
multiple targets

(or land cover types),
which can affect interpretation and
analysis for some applications.

Higher
spatial
resolution

Lower
spatial
resolution


1960 1970 1980 1990 2000

1000






100





10





1





0.1

Polar orbiting meteorological

(

1 km
2
)

Regional
-
scale natural resources


(


1

Ha)

Commercial
satellites (

1 m
2
)

Geostationary satellites

(

5

km
2
)

Local
-
scale natural
resources

(


0.1

Ha)

Pixel size (m)

Trends in
Improving Spatial Resolution

Resolutions of a Digital Image

2.
Spectral Resolution



the number of spectral bands and their
wavelength intervals (or spectral bandwidths) in the EM
spectrum.


Lower resolution
spectral bands

(100 nm & 300 nm [NIR])

Higher resolution
spectral bands

(< 100 nm & 150 nm [NIR])

Certain spectral regions (or bands) are
optimum for obtaining information about
specific characteristics of objects or cover
types
(e.g., vegetation, soils, or water).


Spectral bands are selected to maximize the
contrast between the object of interest
(e.g.,
vegetation
) and its background
(e.g., soils).

Spectral band
locations and
widths for

Landsat Enhanced
Thematic Mapper
(ETM+)

Spectral band location and bandwidth can impact the discrimination
of
cover types and their conditions.


A single,
wide
spectral band
does not discriminate between healthy and
stressed
vegetation,
since the
band value aggregated across multiple spectral regions is
similar, while this
distinction is easily made
from multiple, narrower spectral bands positioned in several parts
of the EM spectrum (“multi
-
spectral” observation).

0

20

Green vegetation

Stressed
vegetation

Reflectance
(%)

0

20

0.4

0.5

0.6

0.7 µm

One band

Three bands

0

20

0.4

0.5

0.6

0.7 µm

0.4

0.5

0.6

0.7 µm

Reflectance(%)

Spectral Resolution


Spectral Resolution


Landsat Multi
-
Spectral
Scanner (MSS) Spectral Bands



Band 1: 0.5 to 0.6
m
m (green)


Band
2: 0.6
to
0.7
m
m
(red)


Band
3: 0.7
to
0.8
m
m
(NIR)


Band
4: 0.8
to
1.1
m
m
(NIR)


Multi
-
spectral and Hyperspectral Data


Multi
-
spectral



multiple bands of
reflected and emitted radiation
across the EM spectrum.


Ex
.
-

Landsat Multispectral Scanner (MSS) = 4 bands


Landsat Thematic Mapper (TM) = 7 bands



SPOT = 3 bands


MODIS = 36 bands


Hyperspectral



very high spectral
resolution of simultaneously
collected images of a large number
of spectral bands (typically 100s)


Ex
.


AISA Eagle = 512 bands


AVIRIS = 224 bands

Resolutions of a Digital Image

3
.
Radiometric Resolution



sensitivity of the sensor to
differences in signal strength as it records the refelcted or
emitted energy from a target (or surface). Defines the
number
of discriminable signal levels
.

8
-
bit

(0
-

255)

10
-
bit

(0
-

1023)

16
-
bit

(0


65,536)

0

0

0

7
-
bit

(0
-

127)

0

Detected reflected or emitted radiation at
sensor is normally quantized during an
analog
-
to
-
digital (A
-
to
-
D) conversion process
to 8, 10, 12 or 16 bits (common bit ranges).

A total of 256 unique values ranging
from 0 to 255 can be represented in

8
-
bit data.

Incident radiation
(photons)

Coded value range by sensor based
on reflected/emitted radiation

Radiometric Resolution


AREA 1:
Bright
areas

AREA 2:
Dark areas

11 bits
:
2048 values

Higher radiometric
resolution

8 bits
:
256 values

Lower radiometric
resolution

(Courtesy Indra Espacio
)

Digital range of numerical values a sensors uses to convert the
detected reflected or emitted radiation from a target to brightness
(or digital) values that comprise the digital image.

(+)
Higher radiometric
resolution increases the
probability that a phenomena
(e.g., leaf area or chlorophyll
content)
may be remotely
sensed more accurately
.


(
-
) Often
requires additional
data processing capabilities
for
computer analysis.


o
R.S. data from most current
sensors have radiometric
resolutions of 8
-
bits or
greater (e.g., Landsat TM = 8
-
bit and MODIS = 16
-
bit).

Resolutions of a Digital Image

4
.
Temporal Resolution



how often a remote sensing system
records imagery over a specific area
(“observation frequency”
or “revisit period”).

18
-
June
-
1998

20
-
July
-
1998

18
-
August
-
1998

Key application question
:
Is the temporal resolution of the data more
frequent than the changes of interest of the application’s target
(e.g.,
phenological/seasonal changes of crops)? If not, key changes or
information related to surface features may be missed.

Temporal Resolution


Revisit cycle of space
-
based r.s.
instruments is
determined by the
orbital characteristics of satellite
, which include:

1)
Altitude (height above the Earth’s surface)

2)
Speed

3)
Declination


a
nd the
IFOV of the sensor
.

Most r.s. instruments used for terrestrial applications
have
a
near
-
polar, sun
-
synchronous orbit
(overpasses
an area at the same local time to be consistent globally).

Sensors on many weather satellites have a
fixed,
geosynchronous (or geostationary) orbit
over the
same area of the Earth’s surface

(allows for
frequent observation of rapidly
-
changing weather
conditions on the order of minutes to hours).

near
-
polar,

sun
-
synchronous
orbit

geosynchronous (or geostationary) orbit

Temporal Resolution


In general, r.s. instruments with
a higher temporal resolution

(more frequent
observations) observe a larger ground area (swath width) with each overpass,
but digital imagery typically have a
lower spatial resolution
.

MODIS

Temporal Resolution:
1 to 2 days

Altitude:
705 km

Swath Width:
2330 km

Spatial Resolution (IFOV):

250, 500 & 1000 meters

Landsat Thematic Mapper (TM)

Temporal Resolution:
14 to 16 days

Altitude:
705 km

Swath Width:
185 km

Spatial Resolution (IFOV):

30
-
meters

VS.

Temporal resolution of
satellite
-
based instruments range from
1 day to twice monthly.
Temporal resolution of
airborne
instruments are usually customizable to a specific data.

Importance of Temporal Resolution

Application Considerations



o
Persistent clouds
limit clear views of the
Earth’s surface (often in tropics or
seasonal


rainy season)


o
Short
-
lived phenomena
(e.g., fire scars,
and floods) that need to be detected


o
Multi
-
temporal observations and
comparison of Earth surface changes
are
needed either within or between years
(e.g., phenology classification, forest die
-
off, or urban growth)

Cloud cover (white) over Kauai, Hawaii in multi
-
spectral
Landsat TM image.

Multi
-
spectral Landsat images of pre
-
fire and active fire
conditions of the High Park fire in Colorado (2012).

Multi
-
spectral Landsat images of urban expansion in
Santiago, Chile (1985 to 2010).

Generating Digital Images


Analog
-
to
-
Digital Conversion


Analog
-
to
-
Digital (A
-
to
-
D) Conversion



conversion of a
continuously varying analog signal

(i.e., incident reflected or
emitted radiation on a detector)
to discrete digital values
(i.e.,
pixel value).






How?:

Sampling the continuous




current (analog signal) at a





uniform interval. All signals





within this interval are represented




as a single average of all variation




in the selected interval.

8
-
bit radiometric resolution

Generating Digital Images


Types of Remote Sensing Scanners

1.
“Whiskbroom” (across
-
track) scanner



physically
moves mirror perpendicular (or across) to the flight
path of the satellite (or plane) to systematically aim
the IFOV over the Earth’s surface.


o
During scanning, the mirror focuses the reflected (or
emitted)
radiation from a specific ground area onto a
discrete set of individual detectors (spectral bands)
sensitive to different regions of the EM spectrum.


o
The voltage produced by each detector is
temporally
sampled at a regular interval

(determines the number
of samples collected along each scan line; pixel
resolution)
and converted to a single digital value


(i.e., pixel BV).

Generating Digital Images


Types of Remote Sensing Scanners

1.
“Pushbroom” (along
-
track) scanners






the sensor’s IFOV is detected simultaneously



by a linear array of detectors

and does not need



a moving mirror
. Array consists of numerous



charge
-
coupled devices (CCDs; detector)



positioned in a line along the array.


o
Each CCD is dedicated to sensing the radiant energy


for multiple spectral bands from a single area of interest



on the ground associated with each element along the linear array.


o
Multi
-
spectral data for each spectral band is compiled when the detected
radiation passes through a dispersing element (prism or dichoric grating) that
separates the EM radiation into separate spectral regions and onto the CCDs
for each array element, which converts the accumulated electrical charge into
a digital value.


o
N
ew ground areas in the direction of the sensor’s along
-
track movement are
subsequently sampled to form new rows of pixel values in the image.

Sensor track
direction

B

G

R

General Types of Sensors:
Passive Sensors

Passive sensors



measure the electromagnetic

radiation derived from external sources; energy

reflected from solar radiation (Sun) or emitted

by the Earth’s surface (thermal).


All sensors collecting data the visible and infrared

(near, middle, and thermal) regions are passive

s
ensors.


Commonly used passive sensors for

agricultural and natural resource applications:



-

Landsat Thematic Mapper (TM )


-

MODIS



-

AVHRR



-

SPOT



-

IKONOS



-

AISA

Passive: Airborne AISA Image

IKONOS

AISA

Landsat TM

MODIS

General Types of Sensors:
Active Sensors

Active sensors



the sensor is the energy source

a
nd detects the reflected energy from the observed

s
urface or target.


Active sensor systems:



1) transmit a pulse(s) of energy towards the



Earth’s surface,




2) the energy interacts with the surface to



produce a backscatter of energy (a “return”



or “echo”)




3) the sensor records both the intensity and



time interval of the backscatter return.

General Types of Sensors:
Active Sensors

Two Main Types of Active Sensors


1)
Active Microwave (RADAR, Radio Detection And Ranging)



longer
microwave wavelengths
(3


25 cm)


2)
LIDAR (Light Detection And Ranging)



relatively shorter
wavelength
laser light

Radar image of
Northern Cascade
Mountains.

Lidar

image of forest

in Colorado Rocky
Mountains.

General Types of Sensors:
Active Sensors

Advantages of Active Sensors

1. Can
penetrate cloud cover

and provide



data about Earth surface conditions under



all types of atmospheric conditions


2.
Acquire information both day and night


3. Contain
valuable information about the



topography and physical structure of the



target/surface

being sensed.


Numerous applications of data:


-

Mapping forest canopy structure



-

Estimate soil moisture


-

Topographic mapping



-

Floodplain mapping


-

Mapping water inundation area and water level height

Image in visible spectrum
with cloud cover (white)

Radar image on same date
with no cloud interference

Manaus, Brazil

Platforms to Collect Remotely Sensed Data

Aircraft

CALMIT

N

Benefits:

o
Collect data for locations



as needed

o
Obtain data when needed

o
Spatial resolution selectable

o
Spectral resolution selectable

o
Configure sensor array



as needed

o
Data available in near
-
real time

o
Control over data collection

Limitations:

o
Costly

o
Not operational, flight missions


have to be planned and there


could be conflicts

o
Access to airborne systems (many


private companies, although



CALMIT/UNL has airborne system)

o
Historical imagery limited or


non
-
existent

Platforms to Collect Remotely Sensed Data

Satellites

o
Wide
-
area coverage

o
Inaccessible locations can be covered

o
Change detection over long periods of time
(e.g., 1972 to present)

o
Aircraft data not available or too expensive

o
Covert operations

Satellite Orbits of Space
-
based Sensors

Sun
-
synchronous, near
-
polar orbit



platform of sensor follows a north
-
south
orbit with a slight inclination (off of polar
north) so they acquire imagery of the ground
area at the same “local” sun time for all parts
of the world.


Ex.


Landsat acquires image data at ~9:45
am local ground time


On each orbit, the sensor on the satellite
images a ‘
swath
’ of the Earth’s surface with
the swath width dependent on the altitude
of the platform. The higher the altitude, the
wider the swath (or ground area that can be
imaged but at a decreasing pixel spatial
resolution.



swath

Satellite Orbits of Space
-
based Sensors

Sun
-
synchronous, near
-
polar
orbit
(continued)


On each orbit (or overpass), a new
ground area is imaged.



Revisit time



time interval between
successive overpass over the same
ground area.



Ex.


16 days between Landsat overpasses



over the same location.


Nadir



the ground location directly
below the satellite (“off nadir” for
other locations).

http://earthnow.usgs.gov/earthnow_app.html?sessionId=049be0ec414b96e18eb2e7ed4c2ec3522457

Satellite Orbits of Space
-
based Sensors

Geostationary orbit



sensors on space
-
based platforms at very high altitudes
(e.g., 36,000 km) that image (or view) the
same portion of the Earth at all times by
revolving (east
-
west direction) at a speed
that matches the Earth’s rotation.


Low spatial resolution images because of
sensors high altitude and large IFOV
(thousands of km
2
).


Commonly used for meteorological
applications related to the monitoring of
the weather.

NOAA GOES satellite image

IFOV

Proximal
(
“In
-
Situ” or “Close
-
Range
”) Sensing



sensor
located in close proximity to a target

and is typically a non
-
imaging
system, which
outputs a multi
-
spectral profile for a given point
(as
compared to an area) instead of a digital image.




Proximal Sensing Data Output



Multi
-
spectral Profile

On Land

On The Water

In The Water

o
Chlorophyll Content

o
Accessory Pigments

o
Leaf Temperature

o
Discrimination of
Plant Species

o
Water Quality Issues

o
Suspended Sediment

o
Algae

o
Wetland Vegetation

o
Coral Reefs

o
Aquatic Plants

Controlled, precise measurement by
proximal sensing

Multi
-
spectral profile output from proximal sensing

Mapping of chlorophyll
concentration in water using
airborne remote sensing data

o
Basic research step that leads to good science (“knowledge discovery”)


o
Quantitative analyses


o
Control troublesome variables affecting traditional remote sensing
(solar and sensor view angle and atmospheric conditions)


o
Understanding gained from proximal remote sensing can form the
basis for remote sensing
applications
and validate their results


Digital Image vs. Aerial Photography

They Are Not the Same

Photographs

are made by
using light to expose film
.


Film contains emulsion layer(s) comprised of silver
halide crystals sensitive to different wavelengths
of
the electromagnetic spectrum are used to record
image in photograph.
Photographs are not digital
,
but can be converted to digital format through
digitization process.


Digital images
are recorded by
converting

the incident radiation received by a detector

sensitive to a specific spectral region to a

digital value assigned to pixel
comprising

an array of pixels (raster) to form the image

with separate bands for each spectral region.

Emulsion layers of natural color film.

Types of Aerial Photos

Black & White Aerial Photo

Natural (or True Color)
Aerial Photo

Color
-
Infrared (CIR)


Aerial Photo

NIR (and blue) sensitive emulsion


Green (and blue) sensitive emulsion


Red (and blue) sensitive emulsion


Support material

Generalized cross
-
section of film emulsions

CIR film is usually exposed through a ‘yellow filter’
that allows green, red, and NIR light to expose the film.

Several types of aerial photos can be and have been used over many years to
assess agricultural, environmental, and natural resource conditions.

DIGITAL
PROCESSING

Energy Source

Receiving System

Application and

End
-
users

Earth
´
s cover



Atmosphere

VISUAL INTERPRETATION

Remote Sensing

Platform

Image data is transferred to Earth
from the satellite using telemetry.
For airborne sensors, the data is
usually on removed hard drive for
data transfer.

Upcoming lectures will focus on
specific remote sensing
instruments and basic processing
and analysis techniques needed to
apply digital image data for a
specific use.