Transcript Remote sensing for hydrology

Remote sensing and Hydrology
Remote sensing:
-Measuring environmental variables without
any direct contact with a target
-Measuring strength of electromagnetic
radiation
-Extraction of valuable information from the
remote sensing data uses mathematically
and statistically based algorithms.
• Understand EM radiative transfer
•Understand sensor characteristics
resolution, orbit, etc.
Electromagnetic energy:
EM wave travel through vacuum at speed of light (c = 3 x 108
m/s). There are two field – electric field and magnetic field –
intersect at right angle. Both vectors are perpendicular to the
direction of wave (wave model)
Wavelength and frequency:
Frequency
ν 
c
λ
Where c = speed of light
(3.0 x 108 m/s)
λ = wavelength
Longer wavelength has higher frequency
Electromagnetic spectrum:
The Sun, earth or any objects emit a continuous spectrum
of energy from gamma rays to radio waves.
Satellite sensors measure EM radiation from visible through
microwave range
Strength of energy emitted
depends on physical body
temperature (-> blackbody
radiation curve).
• Stefan-Boltzmann law
-> Determine total energy, f(T)
• Wein’s displacement law
-> Determine dominant λ
Measure of EM radiation
Radiant flux (Φλ) : energy per unit time, unit = [W]
Radiant flux density (Φλ/A) : unit = [W/m2]
Irradiance: incident radiant flux upon a unit area
Exitance: radiant flux leaving from a unit area
Radiance (Lλ) : Irradiance from a certain direction (θ), unit = [W/m2/sr]
Radiation budget equation
the total amount of incident radiant flux in specific wavelengths
incident (Φi) must be sum of radiant flux reflected from the
surface (Φreflected), the amount of radiant flux absorbed by the
surface (Φabsorbed), and the amount of radiant flux transmitted
through the surface (Φtransmitted):
Φi  Φreflected  Φabsorbed  Φtransmitted
incident
reflection
absorption
transmission
Hemispherical Reflectance, Absorptance, and Transmittance
Divide both side of radiation budget equation by incident radiance
1  ρreflected  ρabsorbed  ρtransmitted
Absorptance (emissivity) ρabsorbed 
Absorptance = emissivity (Kirchhoffs law)
Transmittance
Reflectance
Φabsorbed
Φi
ρtransmitted 
ρreflected 
Φtransmitted
Φi
Φreflected
Φi
Reflectance is often used for remote sensing analysis
All depend on wavelength and materials
Reflectance
Scattering
Redirection of EM radiation by hitting small particles
(typically in the atmosphere)
Three types of scattering:
Function of particle size (gas molecule,
water vapor) relative to wavelength
Rayleigh scattering
Particle size is smaller than wavelength
Scattering amount proportional to λ-4
Mie scattering
Particle size roughly equal to wavelength
Scattering amount proportional to λ-1
Nonselective scattering
Particle size is ~10 times larger than λ
Scattering amount not function of λ
For atmosphere
Remote sensing sensor
Active vs. Passive
Active
• EM Energy is emitted by a sensor toward target
• Measure energy reflected by a target
e.g. radar
Passive
• Measure EM energy emitted by earth or sun
e.g. satellite sensors
Some terminology
Instantaneous field of view (IFOV):
The solid angle over which a measurement is made at any
instance. Given the sensor altitude and IFOV, spatial resolutions
(linear distance) is determined
Swath width
Width of the strip that can be scanned by the sensor.
Nadir
Point on the earth just underneath the sensor
A= IFOV
B= pixel size
nadir
C= altitude
swath
Source: http://ccrs.nrcan.gc.ca/
Satellite orbit
Polar orbit vs. Equatorial orbit
A polar orbit is 90 degree angle of inclination to the equator
(passing north and south poles), whereas an equatorial orbit is
zero degree angle of inclination to equator.
Sun-synchronous (polar orbit)
A special case of polar orbit. Platform pass the same location at
the (roughly) same local time.
Geostationary orbit (equatorial orbit)
A special case of equatorial orbit. Satellite rotate at the same
speed of earth rotation. A satellite appears to be still at the sky all
the time. A satellite altitude is very high (35850 km)
More info -> http://www.rap.ucar.edu/~djohnson/satellite/coverage.html
Polar orbit satellite
One rotation
Rotations
per day
Advantage is daily global coverage
There are ascending path and descending path
Geostationary
Top view
Side view
Need several satellites to cover the entire earth
Geostationary vs. Polar Orbit
G
P
Altitude
High
Low
Speed
Slow
fast
IFOV
Small
large
Sensor resolution
Spatial – the size of field of view (pixel size)
Spectral – range of EM spectrum each band of sensor detects
Temporal – frequency of measurements at a certain location
Radiometric – sensitivity of a sensor to difference in EM energy strength
(recording resolution of sensor)
Radiometric: a sensor records EM energy as brightness value (integer)
2-bit
0
8-bit
0
9-bit
0
Conversion from binary to decimal for 2-bit
00 = 0x21 +0x20 = 0
01 = 0x21 +1x20 = 1
10 = 1x21 +0x20 = 2
11 = 1x21 +1x20 = 3
3
255
511
Sensor resolution
radiometric
B ri g htness val ue
rang e
(typica lly 8 bit)
255
white
spatial
spatial
Lines or
rows (i )
3
4
18
Co lumns ( j)
2
3
4
5
1
1
2
10
15
17
20
15
16
18
18
20
21
22
20
22
17
24
23
21
1
2
22
25
B a nds (k )
3 spectral
127
gray
4
0
X axis
blac k
Pictu re elemen t (pi xel) at lo cat io n
Lin e 4, Col umn 4 , i n Ban d 1 h as a
Bri gh tn ess Val ue o f 2 4 , i.e.,B V4 ,4 ,1 = 2 4 .
Ass o ci ated
g ra y-sca le
Remote sensing – sensor (visible-thermal)
Landsat TM (Thematic Mapper )
Band No.
Wavelength range (μm)
Ground IFOV (m)
1
0.45–0.53 (visible-blue)
30
2
0.52–0.60 (visible-green)
30
3
0.63–0.69 (visible-red)
30
4
0.76–0.90 (Near infrared)
30
5
1.55–1.75 (Near infrared)
30
6
10.40–12.50 (Thermal)
120
7
2.08–2.35 (Mid infrared)
30
Platform = Landsat 4, 5 (sun-synchronous orbit)
Swath width = 185 km
16 day repeat cycle
More info -> http://landsat.usgs.gov/index.php
Remote sensing - sensor (visible-thermal)
Landsat ETM+ (Enhanced Thematic Mapper )
Band No.
Wavelength range (μm)
Ground IFOV (m)
1
0.45–0.515 (visible-blue)
30
2
0.525–0.605 (visible-green)
30
3
0.63–0.69 (visible-red)
30
4
0.75–0.90 (Near Infrared)
30
5
1.55–1.75 (Near Infrared)
30
6
10.40–12.50 (Thermal)
60
7
2.09–2.35 (Mid Infrared)
30
8
0.52–0.90 (panchromatic)
15
Platform = Landsat 7 (sun-synchronous orbit)
Swath width = 185 km
16 day repeat cycle
More info -> http://landsat.usgs.gov/index.php
Remote sensing - sensor (visible-thermal)
AVHRR (Advanced Very High Resolution Radiometer)
Band No. Wavelength range (μm) Ground IFOV (km)
1
0.58–0.68
1.09
2
0.725–1.00
1.09
3A
1.58–1.64
1.09
3B
3.55–3.93
1.09
4
10.30–11.30
1.09
5
11.50–12.50
1.09
Platform = NOAA Polar orbiting Environment satellite
Swath width = 2400 km
Long history since 1979
Daily global coverage (morning and afternoon acquisition)
More info -> http://edcsns17.cr.usgs.gov/1KM/avhrr_sensor.html
Remote sensing - sensor (visible-thermal)
MODIS (Moderate resolution Imaging Spectroradiometer)
There are 36 bands (0.4 - 14.385 μm) visible to thermal
Bands used for land surface
Band No. Wavelength range (μm) Ground IFOV (m)
1
0.620–0.670
250
2
0.841–0.876
250
3
0.459–0.479
500
4
0.545–0.565
500
5
1.230–1.250
500
6
1.628–1.652
500
7
2.105–2.155
500
Platform = EOS Terra and Aqua (Sun-synchronous orbit)
Terra (morning equator-crossing) and Aqua (morning equator-crossing)
Swath width = 2330 km
More info -> http://modis.gsfc.nasa.gov/about/specifications.php
Remote sensing – sensor (passive microwave)
Can measure precipitation, soil moisture, snowpack volume
(SWE, depth), Sea Surface temperature (SST)
Not affected by cloud (visible sensor is affected by cloud)
Coarse spatial resolution
 Polarization
Electric field component (or magnetic field) of EM energy
can vibrate in any directions perpendicular to the direction
of travel. This vibration direction can also evolve with time
vertical
horizontal
Fixed vibration plane
Rotating Vibration plane
Viewed along the travel direction
Brightness temperature (Tb)
Tb value is usually given for passive mircowave data.
Terrestrial matters are not perfect blackbody (graybody).
Total energy emitted by graybody = blackbody radiation (given by
plank law) times emissivity (0<ε<1)
Tb is given using emissivity (Tb = ε*T where T: actual physical
temperature [K])
Emissivity is function of polarization, frequency, and materials
Rayleigh-Jeans approximation
Plank’s law
E λ T  
2hc 2
λ5
x  hc kλT
1
expx   1
Rayleigh-Jeans approximation -> exp(x) ~ 1+x for longer λ
E λ T  
2c  k  T
λ4
Radiation of graybody is given by
I λ T   ε  E λ T   ε
2c  k  T
λ4

2c  k  Tb
λ4
Remote sensing – sensor (passive microwave)
SSM/I (Special Sensor Microwave Imager)
Frequency (GHz)
Polarization
Ground IFOV (km)
19 λ=15.8 mm
Horizontal
25
19
Vertical
25
22 λ=13.6 mm
Vertical
25
37 λ= 8.8 mm
Horizontal
25
37
Vertical
25
85 λ= 3.5 mm
Horizontal
12.5
85
Vertical
12.5
Platform = Defense Meteorological Satellite Program (DMSP)
sun-synchronous orbit
Swath width = 1394 km
Daily global coverage (morning and afternoon acquisition)
More info -> http://nsidc.org/data/docs/daac/nsidc0032_ssmi_ease_tbs.gd.html
Remote sensing – sensor (passive microwave)
AMSR (Advanced Microwave Scanning Radiometer)
Frequency (GHz)
Polarization
Ground IFOV (km)
6.925 λ=43.3 mm
H/V
56
10.65 λ=28.2 mm
H/V
38
18.7 λ=16.0 mm
H/V
21
23.8 λ=12.6 mm
H/V
24
36.5
λ= 8.2 mm
H/V
12
89.0
λ= 3.4 mm
H/V
5.4
Platform = EOS (Earth Observing System) Aqua
Swath width = 1445 km
Daily global coverage (morning and afternoon acquisition)
More info -> http://www.ghcc.msfc.nasa.gov/AMSR/index.html
Application for snow measurement
Use visible – infrared sensors, passive microwave sensor,
depending on what needs to be measured
•Snow cover area (SCA)
Pixel level (Snow / no snow per pixel)
Subpixel level (percentage of SCA over pixel)
•Physical properties of snowpack
Albedo
Grain size
Depth (SWE)
Only estimate of depth (SWE) requires passive microwave data
SCA algorithm (Normalized difference snow index)
To discriminate between Snow and cloud
Source: NOAA NOHRSC
For Landsat TM
NDSI 
TMband2  TMband5
TMband2  TMband5
Use reflectance
Snow if NDSI >0.4
No snow, otherwise
For MODIS
NDSI 
MODISband 4  MODISband 6
MODISband 4  MODISband 6
Snow if NDSI > 0.4 &
Reflectance (band 2) > 11%
No snow, otherwise
TM band2
TM band5
MODIS band4
MODIS band6
SCA algorithm (subpixel level SCA mapping)
Linear spectral mixture analysis
Reflectance measured at each band is a linear combination of
reflectance from individual surface (endmembers) such as snow,
rock, or vege
M
Rλ   Fi  Rλ,i  ε λ
i 1
Rλ: reflectance measured at band of wavelength λ
Rλ,i: reflectance of endmember, i, for band of wavelength λ
Fi : the fraction of endmember, i, over the pixel
M: the number of endmenber
ελ : residual error at wavelength λ
Find F for each endmember with numerical scheme that
minimizes the sum of error
Use multispectral sensors (MODIS, AVHRR, Landsat TM) or
hyperspectral sensors (better because of more bands)
Subpixel level SCA mapping
Binary SCA mapping
Source: Dozier, J., and T. H. Painter, Multispectral and hyperspectral remote sensing of
alpine snow properties, Annual Review of Earth and Planetary Sciences, 32, 465-494
SWE (or snow depth) algorithm
Require passive microwave data because EM radiation from shorter
wavelength (visible – infrared sensors) cannot penetrate full depth of
snowpack, but microwave does.
Tb measured over the snow cover is “cold” compared to bare
ground because snow grains scatters microwave radiation (Mie
scattering)
Algorithm to extract SWE from Tb data set is under
development
Text for remote sensing and useful online
 John R. Jensen, Remote Sensing of the Environment:
http://www.cas.sc.edu/geog/rsbook/Lectures/Rse/index.html
 NASA remote sensing tutorial: http://rst.gsfc.nasa.gov/
 Natural resources Canada, Earth Sciences Sectors:
http://ccrs.nrcan.gc.ca/resource/tutor/fundam/index_e.php
Article for remote sensing for hydrology
Engman, T, E. Recent advances in remote sensing in hydrology,
Reviews of Geophysics, VOL. 33, NO. S1, 967-976, 1995.
- general overview of remote sensing application to hydrology, no
math, a little old
http://www.agu.org/revgeophys/engman00/engman00.html