remote sensing of snow cover
DESCRIPTION
Remote Sensing of Snow Cover. with slides from Jeff Dozier, Tom Painter. Topics in Remote Sensing of Snow. Optics of Snow and Ice Remote Sensing Principles Applications Operational Remote Sensing. Gamma Rays X rays Ultra-violet(UV) Visible (400 - 700nm) Near Infrared (NIR) - PowerPoint PPT PresentationTRANSCRIPT
Remote Sensing of Snow Cover
with slides from Jeff Dozier, Tom Painter
Topics in Remote Sensing of Snow
• Optics of Snow and Ice• Remote Sensing Principles• Applications • Operational Remote Sensing
The EM Spectrum10-1nm 1 nm 10-2m 10-1m 1 m 10 m 100 m 1 mm 1 cm 10 cm 1 m 102m
Gam
ma
Ray
s
X r
ays
Ultr
a-vi
olet
(UV
)
Vis
ible
(40
0 -
700n
m)
Nea
r In
frar
ed (
NIR
)
Infr
ared
(IR
)
Mic
row
aves
Wea
ther
rad
ar
Tel
evis
ion,
FM
rad
io
Sho
rt w
ave
radi
o
Vio
let
Blu
eG
ree
nY
ell
ow
Ora
ng
eR
ed
EM Wavelengths for Snow
• Snow on the ground– Visible, near infrared, infrared– Microwave
• Falling snow– Long microwave, i.e., weather radar
• K ( = 1cm)• X ( = 3 cm)• C ( = 5 cm)• S ( = 10 cm)
General reflectance curves
from Klein, Hall and Riggs, 1998: Hydrological Processes, 12, 1723 - 1744 with sources from Clark et al. (1993); Salisbury and D'Aria (1992, 1994); Salisbury et al. (1994)
Snow Spectral Reflectance
0
20
40
60
80
100
0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4
refl
ec
tan
ce
(%
)
0.05 mm0.2 mm0.5 mm1.0 mm
wavelength (m)
Different Impacts in Different Regions of the Spectrum
Visible, near-infrared, and infrared
• Independent scattering
• Weak polarization
– Scalar radiative transfer
• Penetration near surface only
– ~½ m in blue, few mm in NIR and IR
• Small dielectric contrast between ice and water
Microwave and millimeter wavelength
• Extinction per unit volume
• Polarized signal
– Vector radiative transfer
• Large penetration in dry snow, many m
– Effects of microstructure and stratigraphy
– Small penetration in wet snow
• Large dielectric contrast between ice and water
Visible, Near IR, IR
Mapping of snow extent
• Subpixel problem– “Snow mapping” should estimate fraction of pixel
covered
• Cloud cover– Visible/near-infrared sensors cannot see through
clouds– Active microwave can, at resolution consistent
with topography
Landsat Thematic Mapper (TM)
• 30 m spatial resolution
• 185 km FOV• Spectral resolution
1. 0.45-0.52 μm2. 0.52-0.60 μm3. 0.63-0.69 μm4. 0.76-0.90 μm5. 1.55-1.75 μm6. 10.4-12.5 μm7. 2.08-2.35 μm
• 16 day repeat pass
AVIRIS spectraAVIRIS spectra
0
20
40
60
80
100
0.3 0.8 1.3 1.8 2.3wavelength (m)
refl
ec
tan
ce
(%
)
snow
vegetation
rock
Spectra of Mixed PixelsSpectra of Mixed Pixels
0
20
40
60
80
100
0.3 0.8 1.3 1.8 2.3wavelength (m)
refl
ec
tan
ce
(%
)
snow
vegetation
rock
equal snow-veg-rock
80% snow, 10% veg, 10% rock
20% snow, 50% veg, 30% rock
• Assuming linear mixing, the spectrum of a pixel is the area-weighted average of the spectra of the “end-members”
• For all wavelengths ,
• Solve for fn
Analysis of Mixed PixelsAnalysis of Mixed Pixels
R r fn nn
N
1
Subpixel Resolution Snow Mapping from Landsat Thematic Mapper
Subpixel Resolution Snow Mapping from Landsat Thematic Mapper
Sept 2, 1993(snow in cirques only)
Feb 9, 1994(after big winter storm)
Apr 14, 1994(snow line 2400-3000 m)
(Rosenthal & Dozier, Water Resour. Res., 1996)
Subpixel Resolution Snow Mapping from AVHRR
Subpixel Resolution Snow Mapping from AVHRR
May 26, 1995
(AVHRR has 1.1 km spatial resolution, 5 spectral bands)
AVHRR Fractional SCA Algorithm
1
2
3
4
5
AVHRR (HRPT FORMAT)Pre-Processed at UCSB[NOAA-12,14,16]
Snow Map Algorithm Output: Mixed clouds, high reflective bare ground, and Sub-pixel snow cover
AVHRR Bands
Geographic Mask
Thermal Mask
Masked Fractional SCA Map
Composite Cloud Mask
Build Cloud Masks using several
spectral-based tests
Execute Atmospheric Corrections,
Conversion to engineering units
Execute Sub-pixel snow cover algorithm
using reflectance Bands 1,2,3
Application of Cloud, Thermal, and Geographic masks to raw
AVTREE output
Build Thermal Mask
Scene Evaluation: Degree of Cloud Cover
over Study Basins
Subpixel Resolution Snow Mapping from AVIRIS
Subpixel Resolution Snow Mapping from AVIRIS
(Painter et al., Remote Sens. Environ., 1998)
EOS Terra MODIS
•Image Earth’s surface every 1 to 2 days
•36 spectral bands covering VIS, NIR, thermal
•1 km spatial resolution (29 bands)
•500 m spatial resolution (5 bands)
•250 m spatial resolution (2 bands)
•2330 km swath
Discrimination between Snow and Glacier Ice, Ötztal Alps
Discrimination between Snow and Glacier Ice, Ötztal Alps
Landsat TM, Aug 24, 1989 snow ice rock/veg
Snow Water EquivalentSnow Water Equivalent
• SWE is usually more relevant than SCA, especially for alpine terrain
• Gamma radiation is successful over flat terrain
• Passive and active microwave are used• Density, wetness, layers, etc. and vegetation
affect radar signal, making problem more difficult
SWE from Gamma
• There is a natural emission of Gamma from the soil (water and soil matrix)
• Measurement of Gamma to estimate soil moisture
• Difference in winter Gamma measurement and pre-snow measurement – extinction of Gamma yields SWE
• PROBLEM: currently only Airborne measurements (NOAA-NOHRSC)
Microwave Wavelengths
Frequency Variation for Dialectric Function and Extinction Properties
• Variation in dialectric properties of ice and water at microwave wavelengths
• Different albedo and penetration depth for wet vs. dry snow, varying with microwave wavelength
• NOTE: typically satellite microwave radiation defined by its frequency (and not wavelength)
Modeling electromagnetic scattering and absorption
Soil
(1) (2) (3) (4) (5) (6)
Snow
SWE and Other Properties derived from SIR-C/X-SAR
Particle radiusSIR-C/X-SAR Snow density Snow depth
Est
imat
ed
Ground measurements
Snowdensity
Snow depthin cm
Grain radiusin mm
Passive Microwave SWE Estimates
• Microwave response affected by:– Liquid water content, crystal size and shape, depth
and SWE, stratification, snow surface roughness, density, temperature, soil state, moisture, roughness, vegetation cover
• Ratio of different wavelengths– Vertically polarized brightness temperature, TB,
gradient
– Single frequency vertical polarized TB
BTVdcSWE GHz 37
19/ GHz 18 GHz 37 BB TVTVbaSWE
Passive Microwave SWE Estimates
• Advantages:– Daily overpass (SSM/I, Nimbus-7 SMMR)– Large coverage areas– Long time series (eg. Cosmos 243 - Russia 1968)– See through clouds, no dependence on the sun
(unlike visible or near IR)
• Disadvantages– Large pixel size (12.5 – 25 km)– Still problems with vegetation– Maximum SWE & limitations with wet snow
Passive Microwave SWE Products
Active Microwave Snow Detection
• Has been used to estimate binary SCA at 15 - 30 m resolution as compared to air photos
• Advantages:– High resolution– Detection characteristics
• Disadvantages:– Repeat of 16 days & narrow Swath width, as per TM– Commercial sensor: ERS-I/II (?), RADARSAT
Active Microwave SWE Estimation
• Snow cover characteristics influence underlying soil temperature, this affects the dielectric constant of soil
• Backscatter from soil influenced by dielectric constant and by soil frost penetration depth
• Snow cover insulation properties influence backscatter
from Bernier et al., 1999: Hydrol. Proc. 13: 3041-3051
Active Microwave SWE EstimationRSWE s bmR o
row
Th
erm
al s
no
w r
esis
tan
ce
(R i
n o
Cm
3s/
J)
Backscattering ratio (w
o - ro in dB)
SW
E /
R
Mean snow density (s in km/m3)
Problem: Maximum SWE detectable in order of 400 mm
from Bernier et al., 1999: Hydrol. Proc. 13: 3041-3051
Weather Radar for Snowfall
• Ground-based NEXRAD system covers most of the conterminous US, except some alpine areas
• Snowfall estimation improves with time of accumulation, not necessarily required for individual storm events like rainfall
• Variation in attenuation due to particle shape, wet snow, melting snow
• General problems with weather radar
Weather Radar vs. Gauge Accumulation
from Fassnacht et al., 2001: J. Hydrol. 254: 148-168
0
50
100
150
200
250
300
0 28 56 84 112 140
time increment (days)
pe
rce
nta
ge
ab
so
lute
dif
fere
nc
e(r
ad
ar
- g
au
ge
)
Particle Characteristics Considerations
from Fassnacht et al., 2001: J. Hydrol. 254: 148-168
0
50
100
150
200
250
0 50 100 150monthly gauge accumulation (mm)
mo
nth
ly r
ad
ar
ac
cu
mu
lati
on
(m
m) Wormwood
Greenoch
Euclid1:1 line
0
50
100
150
200
250
0 50 100 150monthly gauge accumulation (mm)
mo
nth
ly r
ad
ar
ac
cu
mu
lati
on
(m
m)
0
50
100
150
200
250
0 50 100 150monthly gauge accumulation (mm)
mo
nth
ly r
ad
ar
ac
cu
mu
lati
on
(m
m)
0
50
100
150
200
250
0 50 100 150monthly gauge accumulation (mm)
mo
nth
ly r
ad
ar
ac
cu
mu
lati
on
(m
m)
Scaling removed
Mixed precipitationRaw
mixed precip + particle shape
Research / Operational Products
• Snow-covered area– Fractional SCA with Landsat or AVHRR (UAz RESAC)– With AVIRIS, also get albedo– Binary SCA currently from MODIS, VIIRS (NPOESS)
• Snow-water equivalent– L-band dual polarization + C- and X-band– Daily SSM/I over the Midwest and Prairies
• Snow wetness– Near surface with AVIRIS– Within 2% with C-band dual polarization