Satellite-Based Ocean Surface Turbulent Fluxes - Satellite-based Applications on Climate Change - page 164

(25-50 km). Passive microwave instruments for SST retrievals include SMMR, TMI,
and AMSR-E. SST observations have been greatly improved by combination of
multiple infrared and microwave sensors (Donlon et al. 2007 ).
11.3.3 Surface-Air Temperature and Humidity
Determining T a and Q a from satellites is difficult and remains challenging.
Satellite sounders such as the Atmospheric Infrared Sounder (AIRS) and the
Advanced Microwave Sounding Unit (AMSU) provide profiles of air temperature
and humidity, but do not adequately resolve the boundary layer. Microwave sensor
channels are mostly sensitive to the total precipitable water ( W ) or water vapor
within a thin layer, typically 500 m, close to the sea surface (WB) (e.g., Liu 1986 ;
Schulz et al. 1993 ).
Climatologically, there is a strong relationship between the total columnar
precipitable water and the surface-air humidity (Liu 1986 ). This relationship is
based on the strong coupling and feedbacks between the surface and the atmosphere
above the boundary layer. However, for mesoscale systems, such as fronts and cold
air outbreaks, this relation is inadequate for describing the surface property using
columnar measurements. Application of the climatological relationship to SSM/I
data produced large spatial bias in LHF estimates (Esbensen et al. 1993 ; Schlussel
et al. 1995 ; Schulz et al. 1997 ). Schulz et al. ( 1993 ), Schlussel et al. ( 1995 ), and
Bentamy et al. ( 2003 ) improved Q a retrieval by establishing a relationship between
Q a and WB which is estimated from SSM/I. Chou et al. ( 1995 ) classified sounding
data observed during the First Global Atmospheric Research Program (GARP)
Global Experiment (FGGE) and found that the first two Empirical Orthogonal
Functions (EOFs) of the vertical moisture profile can explain most of the variance.
By solving two simultaneous equations involving the W and WB, Q a is estimated.
This technique incorporates additional information about the vertical structure of
the atmosphere that has not been considered in previous retrievals.
Early methods (e.g., Liu 1988 ; Kubota and Shikauchi 1995 ) determined T a from
satellite observations by using a specified value of relative humidity (usually 80%)
with a known Q a via the Clausius-Clapeyron relationship. The spatial and temporal
variations of relative humidity are as yet relatively unexplored. Subsequent studies
by Jourdan and Gautier ( 1995 ) and Konda et al. ( 1996 ) used additional variables
such as W , SST, and wind speed to estimate T a . More recent improvement of Q a and
T a estimates involves the use of robust techniques such as artificial neural networks
(Jones et al. 1999 ; Roberts et al. 2010 ) and genetic algorithms (Singh et al. 2005 ,
2006 ), or from a combination of multisensor observations (Jackson et al. 2006 ,
2009 ; Jackson and Wick 2010 ).