Passive Microwave Remote Sensing Lecture 10 Nov 06, 2007

Principals While dominate wavelength of Earth is 9.7 um (thermal), a continuum of energy is emitted from Earth to the atmosphere. In fact, the Earth passively emits a steady stream of microwave energy as well, though it is relatively weak in intensity due to its long wavelength. The spatial resolution usually low (kms) since the weak signal. A suit of radiometers can record it. They measure the brightness temperature of the terrain or the atmosphere. This is much like the thermal infrared radiometer for temperature. A matrix of brightness temperature values can then be used to construct a passive microwave image. To measure soil moisture, precipitation, ice water content, sea-surface temperature, snow-ice temperature, and etc.

Rayleigh-Jeans approximation of Planck’s law Thermal infrared domain (Planck’s law): Microwave domain (Rayleigh-Jeans approximation): Recall Let We have We have Unit is Wm-2Hz

For a Lambertian surface, the surface brightness radiation B(v,T), The really useful simplification involves emissivity and brightness temperature: Unit is W•m-2•Hz•sr In comparison with thermal infrared: (TB)4 = ελ (T)4

Some important passive microwave radiometers Special Sensor Mirowave/Imager (SSM/I) It was onboard the Defense Meterorological Satellite Program (DMSP) since 1987 It measure the microwave brightness temperatures of atmosphere, ocean, and terrain at 19.35, 22.23, 37, and 85.5 GHz. TRMM microwave imager (TMI) It is based on SSM/I, and added one more frequency of 10.7 GHz.

AMSR-E Advanced Microwave Scanning Radiometer – EOS It observes atmospheric, land, oceanic, and cryospheric parameters, including precipitation, sea surface temperatures, ice concentrations, snow water equivalent, surface wetness, wind speed, atmospheric cloud water, and water vapor. At the AMSR-E low-frequency channels, the atmosphere is relatively transparent, and the polarization and spectral characteristics of the received microwave radiation are dominated by emission and scattering at the Earth surface. Over land, the emission and scattering depend primarily on the water content of the soil, the surface roughness and topography, the surface temperature, and the vegetation cover. The surface brightness T (TB ) tend to increase with frequency due to the absorptive effects of water in soil and vegetation that also increase with frequency. However, as the frequency increase, scattering effects from the surface and vegetation also increase, acting as a factor to reduce the TB

AMSR-E Najoku et al. 2005

Example1: Snow depth or snow water equivalent (SWE) The microwave brightness temperature emitted from a snow cover is related to the snow mass which can be represented by the combined snow density and depth, or the SWE (a hydrological quantity that is obtained from the product of snow depth and density). ∆Tb = Tb19V-Tb37V

Kelly et al. 2003

3. Study Area (1)

Impact of snow density (4)-mean SD Snow density = 0.4 g/cm3 Multi-snow density Xianwei, Xie, and Liang 2006

Results: AMSR-E vs ground- SD at individual stations (snow density = 0 Results: AMSR-E vs ground- SD at individual stations (snow density = 0.4 g/cm3)

Results: AMSR-E vs ground- SD at individual stations (snow density = 0 Results: AMSR-E vs ground- SD at individual stations (snow density = 0.4 g/cm3)

Results: Annual change of SWE in YWR

Antarctic sea ice

Mean SD appears relatively consistent for most of the year with large increase in summer months. Peaks near the lowest point of area coverage. Mean SD appears to increase from 2003 to 2005 during the peak. Maximum SD values exceed 50-60 cm in most data sets, (outside range of retrievable snow depth for 37GHz) and are likely noise. The effect of maximums on mean SD is expected to be small based on the small number of pixels with values greater than 50 centimeters. Greatest Mean SD occurs in summer where there is the least amount of coverage area. Seasonal variation shows up as hysteresis. Summer peaks in mean SD likely the result of reduced sea ice extent and show a bias toward thicker (remaining) snow accumulations. 2005 summer shows least coverage area and greatest mean SD – approaches 50 centimeter retrieval limit. Minimum area of snow on sea ice occurs at the end of the summer (late February) and maximum in spring (late September) Snow coverage area on sea ice appeared to decrease in late summer from 2003 to 2005 and increase in early spring from 2002 to 2005 Mike and Xie, 2006

Mean SD appears relatively consistent for most of the year with large increase in summer months. Peaks near the lowest point of area coverage. Mean SD appears to increase from 2003 to 2005 during the peak. Maximum SD values exceed 50-60 cm in most data sets, (outside range of retrievable snow depth for 37GHz) and are likely noise. The effect of maximums on mean SD is expected to be small based on the small number of pixels with values greater than 50 centimeters. Maximum SD values exceed 50-60 cm in most data sets, (outside range of retrievable snow depth for 37GHz) and are likely noise Mike and Xie, 2006

Greatest Mean SD occurs in summer where there is the least amount of coverage area. Seasonal variation shows up as hysteresis. Summer peaks in mean SD likely the result of reduced sea ice extent and show a bias toward thicker (remaining) snow accumulations. 2005 summer shows least coverage area and greatest mean SD – approaches 50 centimeter retrieval limit. Mike and Xie, 2006

Snow Volume estimated as product of mean SD and total area of snow cover. Will this change with pixel-by-pixel calculation? Appears to be a reduction is snow volume from 2002 to 2005 in both peak (spring) and low (summer) seasons

Seasonal Comparison of Locations of Max SD Areas, 2002 Max Areas = +2σ 7/20/02 10/20/02 11/18/02 8/20/02 9/24/02 12/20/02

Seasonal Comparison of Locations of Max SD Areas, 2003 1/20/03 4/20/03 7/20/03 10/20/03 2/20/03 5/20/03 8/20/03 11/18/03 Oct 1, 2004 Oct 1, 2005 3/20/03 6/20/03 9/20/03 12/20/03

Seasonal Comparison of Locations of Max SD Areas, 2004 1/20/04 4/20/04 7/22/04 10/20/04 11/17/04 2/20/04 5/20/04 8/20/04 3/20/04 6/19/04 9/17/04 12/20/04

Seasonal Comparison of Locations of Max SD Areas, 2005 4/20/05 7/20/05 10/20/05 1/20/05 2/20/05 5/20/05 8/20/05 11/16/05 3/20/05 6/20/05 9/20/05 12/20/05

Example2: Radio-frequency interference contaminate the 6. 9 and 10 Example2: Radio-frequency interference contaminate the 6.9 and 10.7 GHz channels Radio-frequency interference (RFI): the cable television relay, auxiliary broadcasting, mobile. RFI is several orders of magnitude higher than natural thermal emissions and is often directional and can be either continuous or intermittent. Radio-frequency interference (RFI) is an increasingly serious problem for passive and active microwave sensing of the Earth. The 6.9 GHz contamination is mostly in USA, Japan, and the Middle East. The 10.7 GHz contamination is mostly in England, Italy, and Japan RFI contamination compromise the science objectives of sensors that use 6.9 and 10.7 GHz (corresponding to the C-band and X-band in active microwave sensing) over land.

radio-frequency interference (RFI) index (RI)

Li et al. 2004

6.9 GHz contamination Najoku et al. 2005