Transcript Remote Sensing

READING AND MODELLING
Review about mass balance modelling:
Greuell, W., and C. Genthon, 2004: Modelling land-ice surface
mass balance. In Bamber, J.L. and A.J. Payne, eds. Mass
balance of the cryosphere: observations and modelling of
contemporary and future changes. Cambridge University
Press.
Mass balance model that includes sub-surface module:
http://www.phys.uu.nl/%7Egreuell/massbalmodel.html
REMOTE SENSING OF GLACIERS
Karthaus, September 2005
Wouter Greuell
Institute for Marine and Atmospheric Research Utrecht (IMAU)
Utrecht University, the Netherlands
Retrieval of: - surface velocity
- surface topography
- glacier facies
- surface albedo
WAVE LENGTHS AND RETRIEVABLE INFORMATION
Glaciology
Frequency (GHz)
300000
30000
3000
300
30
3
0.3
Types of sensors
microwave sensors (e.g. radar)
short-wave:
emitted by Sun
albedo
glacier facies
surface topography
0.1
1
Types of radiation (physics)
surface
temperature
10
Terms in
meteorology
Retrievable
information
blue: active
green: passive
100
1000
Wave length (µm)
L-band
long-wave:
emitted by
atmosphere
and surface
of the Earth
C-band
microwaves (e.g. radar bands)
thermal infrared
X-band
mid infrared
visible
near infrared
ultraviolet
optical sensors
surface velocity
glacier facies
surface topography
104
105
106
ORBIT (LANDSAT 7)
altitude = 705 km
inclination = 98˚
No data beyond 82 ˚N and S
orbit period = 99 min
ground-track speed ~ 6.7
km/s
crosses equator at 9:45 AM
local time (for optical
sensors)
SCANS
LANDSAT TM (1,
2 and 3)
total field of view: 11.6˚
scan mirror oscillates
once per 33 msec
6 detectors per band i.e.
six contiguous lines
for each mirror
semi-oscillation
COVERAGE
LANDSAT 7
orbit repeated after 16 days
SAR
INTERFEROMETRY
InSAR
1) velocity field
2) topography
SIDE-LOOKING RADAR
(REAL-APERTURE RADAR)
Active
sensor !
Emission of a short pulse: tp = 30 ns
Across-track resolution obtained
by time-resolving the signal
Rr 
ct p
2sin 
if tp = 30 ns, = 35˚ then Rr = 8 m
Along-track resolution is poor
Ra 
H
L cos
if H = 800 km,  = 24 cm, L = 10 m
= 35˚ then Ra = 24 km
SYNTHETIC APERTURE RADAR (SAR)
Increase along track resolution
pulse repetition frequency: 1000 Hz
satellite speed: 6 km s-1
Every 6 m a sample is taken
Therefore, 4000 measurements are taken within
24 km (the original along-track resolution)
Every measurement contains the information from
4000 ground elements of 6 m, but each ground
element is sampled 4000 times ......
computer time-demanding procedure (called
focusing) with complex numbers
PRINCIPLE OF SAR
INTERFEROMETRY
- Use 2 images (A and B) from
repeat orbits (typically no more
than a few 100 m apart = d)
- Use phase ()
- Range (R; distance satellite to
pixel) = n (integer number) *  (6
cm) + /2 * 
- So phase gives some info about
range (but n is unknown!)

- Take difference of phases from
two images for each pixel =
difference in range from two orbits
( but (nA-nB)  is unknown!)
- Make image of phase difference (=
interferogram)
Orbit A
Image A
Orbit B
Image B
d
R A  nA 
B

2
R B  n B 
B

2

pixel i pixel i+1
Contributions to interferometric signal:
- Differences in positions orbits
- Surface topography
- Surface displacements
RAW INTERFEROGRAM
Contours (colours) connecting points
of equal phase difference are called
fringes
THE INTERFEROMETRIC LIMIT
dR = difference
in path length for extreme
ends of one pixel
dR1 = 2 L sin 1
signal from a pixel is the sum of hundreds
of elementary targets
phase from pixel is random
is removed by differencing two signals
from the same pixel
Targets must remain stable between
image acquisitions (e.g. 3 days)
1
Pixel must not stretch or shrink by more
than a fraction of  from one image to the
other
2 L ( sin 1 - sin 2 ) < 
1 pixel
L
orbital separation (d) should be < 1 km
if d too large: incoherence
ALTITUDE OF AMBIGUITY
= shift in altitude of the surface corresponding to a
phase shift of 2π in the interferogram
R
ha 
2dtan 
ha:
R:
 :
:
d:
altitude of ambiguity
range from satellite to target
wavelength
angle of incidence
horizontal separation of trajectories
If interested in topography:
large d
If interested in displacements: small d
d

R
ha
DIGITAL ELEVATION MODEL
glacier free terrain in Alaska
a
b
prior (a) and after (b) removal of orbital effect
Note: phase needs to be unwrapped. Tie points needed!
SEPARATE VELOCITY
FROM TOPOGRAPHIC
FIELD
Images
Option 2:
differential
interferometry:
use two
interferograms and
assume constant
velocity
Option 1: create synthetic
interferogram from known topography
and und subtract this from measured
interferogram
Interferograms
Differential
interferograms
Day 1
altitude of ambiguity
ha1
Day 4

topography (ha1)
+
velocity
Day 16
altitude of ambiguity
ha2
Day 19

topography (ha2)
+
velocity

velocity only

topography
only
because ha1
differs from ha2

velocity only
EXAMPLE
OF
VELOCITY
FIELD
Bagley Ice Field
Alaska
a) Date 1: topography
(h) and velocity (v)
b) Date 2: h + v
c) h only
d) Date 2: v only
ESTIMATE SURFACE
VELOCITY
Limitation!
Calculated velocity = velocity
along line connecting the
target with the satellite
Extra info:
- another interferogram
- assume surface parallel flow
- assume flow along the
surface gradient
- assume flow along valley
walls
DETECTING GLACIER FACIES
GLACIER FACIES
SNOW LINE FROM NEAR-INFRARED IMAGERY
Morteratschgletscher
TM band 4 (800 - 900
nm)
24 June 1999
EFFECTS ON THE RADAR SIGNAL
amplitude!
 = 3 -25 cm
satellite
sensor
incidence
angle (20 - 50Þ)
rough surface
(h > 1 cm):
elements < 1 cm:
elements > 1 cm: do
smooth surface
(h < 1 cm):
by ice (absorption length = 10 m):
by water (absorption length = 5 mm):
THE RESULTING RADAR SIGNAL
Atmosphere
some absorption by clouds with water
droplets, but no scattering
Water in snow or on ice strong absorption
Most glacier surfaces
are rough
All images useful
use winter images
backscattered signal depends on shape
of roughness elements
Volume scattering increases with concentration of large ( > 1 cm ) inhomogeneities
Facies
Large
inhomogeneities
Signal
dry-snow
few
weak
wet snow and percolation
grain clusters and ice lenses
strong
ice (transformed from snow) some air bubbles and cracks
medium
superimposed ice
between strong and medium
more air bubbles
AN EXAMPLE: FACIES ON KONGSVEGEN
(SVALBARD) FROM SAR
1-4
ice
5-6
supimp.
ice
8-9
snow
RADAR ALTIMETRY (PRINCIPLE)
Principle: - emittance of a short (tp = 3 ns) pulse
- detection of the return
- determination of the travel time (Tt)
- calculation of the distance to the surface (H)
Tt c
H
2
RADAR ALTIMETRY: RANGE RESOLUTION
H = 0.5 c tp
H ≈ 0.5 m
RADAR ALTIMETRY: FOOTPRINT
The footprint is “pulselimited” (and not “beam
limited”)

Footprint (x) = diameter of circle when rear front hits surface
H  0.5 x  H  ct p 
2
2
2
From H >> ctp, it follows:

x  8ct pH
x ≈ 2400 m
SLOPE-INDUCED ERROR
R2
R1
platform motion
H2
H1
H1
S2
S1

ce
a
f
r
su
n
i
a
H2
terr
t
n
e
ace
r
f
a
r
p
u
s
ap
n
i
a
r
ter
e
u
tr
S2
S1
Technique works only when slope < 1 degree
CHANGE IN ELEVATION
AT CROSSING POINTS
Only at crossing point of ascending and descending tracks,
because repeat tracks are too far apart (a few km)
ELEVATION CHANGE ANTARCTICA
Period: 1992 - 1996
No orbits beyond
about 81 ˚S
Only measurements
when slope < 1˚
ELEVATION CHANGE WITH AIRBORNE LASER
Same principle as radar altimeter, but:
- flight lines are repeated exactly, leading
to info along entire flight lines
- Footprint ≈ 1 m
- Direction of reflection known with large
accuracy (no problem over steep terrain)
but
- Total length of flight lines limited
Elevation changes Greenland 1997-2003
TOPOGRAPHY FROM RADAR, LASER AND SAR
Radar on board
ENVISAT
Laser on board
ICESat
SAR
interferometry
Wavelength (nm)
2.2 and 9.4 cm
530 and 1060 nm
3 - 24 cm
Spatial resolution
1.7 - 2.4 km
pulse-limited
70 m
beam-limited
80 m
Range resolution
5 - 20 cm
10 cm
5 - 20 m
Can be used if
slope > 1˚
no
yes
yes
Penetrates through
clouds
yes
no
yes
Useful for big ice
sheets
yes
yes
tie points needed
Product after one
flight line
1D
1D
2D
SUM UP
1) Orbits, swath, resolution
2) SAR interferometry for surface velocity and
topography
3) Glacier facies with optical sensors and SAR
4) Altimetry with radar (laser and SAR)
SOME READING
Introduction to remote sensing:
Rees, W.G., 2001: Physical principles of remote sensing. Cambridge University Press,
Cambridge (U.K.), 343 pp.
Review about remote sensing of snow and ice:
König, M., J.-G. Winther and E. Isaksson, 2001: Measuring snow and glacier
properties from satellite. Rev. Geophys., 39 (1), 1-27.
Review about SAR interferometry:
D. Massonnet and K. L. Feigl, 1998: Radar interferometry and its application to
changes in the Earth’s surface. Rev. Geophys., 36 (4), 441-500.
Paper about using SAR interferometry to derive glacier velocity field:
Fatland, D.R. and C.S. Lingle, 1998: Analysis of the 1993-95 Bering Glacier (Alaska)
surge using differential SAR interferometry. J. Glaciol., 44 (148),532-546.
SIGNALS ARE AVERAGED
for ERS-1 over 50 returns
Real frequency: 20 Hz
Distance of info along track = 330 m
RANGE WINDOW
Signal is sampled within short time interval (relative to pulse
repetition time) in order to reduce data volume = range window
Half-power point = retrack point = mean surface elevation
within footprint for Gaussian distribution of slopes
Onboard tracker tries to predict travel
time of next return in order to place
range window correctly
When signal is missed altogether: loss
of lock
Sensor goes into “acquisition mode”:
no data for a few seconds
MEASURED RETURN
SIGNALS
- every signal is mean of 50 returns
- every sixth signal is shown
- flight over margin Greenland ice
sheet
- 40-48: coast
- 106-232: loss of lock in rugged
terrain
- 238-274: ice sheet
PULSE REPETITION FREQUENCY
For ERS-1: 1 kHz
Pulses are 10-3 s apart, compare to pulse duration of 3 ns
ERROR SOURCES
Relevant for changes in ice-elevation measurements:
- Atmospheric
a) dry atmosphere
b) wet atmosphere
c) ionosphere
- Orbit
- Variation in sub-surface properties (from which depth is the signal
reflected?)
- Slope (see next slide)
Note also that changes in snow (ice) density without changes in elevation do
not affect volume, but they do affect mass and sea level
DETERMINATION OF THE SURFACE TEMPERATURE
BLACK-BODY RADIATION
1
Normalized (300 K) radiance
Band of measurement
0.8
275 K
300 K
0.6
0.4
0.2
250 K
0
5
10
15
20
25
Wavelength (µm)
30
35
40
BRIGHTNESS TEMPERATURE
This temperature is called the brightness temperature
Satellite sensors are calibrated on-board with
blackbodies of known temperatures
The real surface temperature is several degrees
centigrade higher than the brightness temperature due to
absorption in the atmosphere
ATMOSPHERIC WINDOWS
AVHRR bands 4 and 5 are situated in the atmospheric window
between 10.3 and 12.5 µm
SPLIT WINDOW AND DUAL VIEW
Difference real surface - brightness temperature varies with:
amount of greenhouse gases along atmospheric path
- concentration of gases (e.g. water vapour)
- surface elevation
If this is unknown:
Ts = a0 + a1 TB1 + a2 TB2
where Ts:
a i:
TBi:
surface temperature
constants
brightness temperatures obtained from different sensors
Split window:
brightness temperatures from two different spectral bands
Dual view:
brightness temperatures from two different angles
Equation optimized by means of measurements or calculations
SAR INTERFEROMETRY
= differencing the phases of two SAR images
phase = range = distance between satellite and ground target
difference in phase = difference in range (between 2 images)
is not absolute, but relative !

= 0
= 
= 1.5 
Contributions to interferometric
signal:
- differences in orbital trajectories
- surface topography
- surface displacements