Satellite Remote Sensing of Evaporation over Sea
437
as seen in figure 16.1. However, the radiation field at 3.7 pm not only contains thermal radiation emitted from the earth and its atmosphere; during day it also contains a substantial part
of scattered solar radiation which hampers the exploitation of radiance mesurements in this
spectral region with respect to sea surface temperature. During night, however, the brightness
temperatures measured in the 3.7 pm region can be used together with measurements at 11
or 12 pm in a similar way described by (16.3), rendering the split-window into a dual-window
technique. Unfortunately, past measurements at 3.7 pm as taken by the AVHRR instrument
were too noisy for a suitable utilization and the dual-window techniques were less frequently
employed than the split-window method. The ATSR that is flown on the ERS-1 satellite carries
a 3.7 pm channel with low noise amplitudes that allows its full utilization for the atmospheric
correction of nocturnal infrared imagery when deriving sea surface temperature.
In some cases the coefficients being used in (16.3) are tuned to distinct regions in order to optimize the temperature retrieval for specific atmospheric situations (e.g. Llewellyn-Jones et aI.,
1984; Minnett, 1990). This regional tuning is only partly successful since even in small areas
the variability of atmospheric water vapour can reach high values. For example, in the tropical
Pacific Ocean day to day variations of the water-vapour columnn by more than 2 gjcm 2 can
be observed in SSMjI water-vapour measurements. Besides this the regionalization of retrieval
schemes can lead to discontinuities in the derived fields at the boundaries between adjacent
regions so that in large-scale applications temperature gradients can be faked. However, with
respect to studies of limited areas the regionalisation of retrieval schemes might be successful.
A further possibility for the improvement of the surface temperature retrieval is the knowledge
of the actual state of the lower atmosphere. This could be obtained from radiosoundings or
via satellite-derived temperature and moisture profiles. Even more advantageous seems the
direct utilization of additional information from other instruments flying on the same satellites.
Case studies for the synergetic use of different radiometers for the atmospheric correction have
shown improved performances when compared to the use of a two- or three-channel technique
(Schliissel et aI., 1987; Wick et aI., 1992). A disadvantage of this combination of instruments
is the difference of the horizontal resolution of the radiometers in use. While AVHRR resolves
an area of 0.95 km 2 at nadir view an infrared sounder like the High-resolution Infrared Radiation Sounder (HIRS) covers 227 km 2 in a pixel near the sub-satellite track. Computationally
expensive interpolations are necessary in order to match one instrument to the other.
16.2.1 Retrieval techniques
Usually, radiative transfer calculations are used for the development of retrieval techniques.
Each retrieval scheme has to be optimized according to the intrument characteristics because
of different noise behaviour and spectral response functions. It has been shown that the utilization of split-window coefficients developed for one radiometer (AVHRRj2 on board NOAA-7)
in the retrieval of To from another radiometer (AVHRRj2 on board NOAA-9) can introduce
systematic errors up to 0.29 K (Minnett, 1990).
Examples of simulated brightness temperature deficits in the 12 pm channel at different viewing angles of the ATSR are shown in figure 16.2 for 289 globally distributed test cases. The
corrections necessary broadly vary with the surface temperature which is roughly correlated
with the atmospheric water-vapour amount. The slant view through the atmosphere strongly
increases the differences. The values vary between -2 and 16 K at nadir view and up to 22 K
437
as seen in figure 16.1. However, the radiation field at 3.7 pm not only contains thermal radiation emitted from the earth and its atmosphere; during day it also contains a substantial part
of scattered solar radiation which hampers the exploitation of radiance mesurements in this
spectral region with respect to sea surface temperature. During night, however, the brightness
temperatures measured in the 3.7 pm region can be used together with measurements at 11
or 12 pm in a similar way described by (16.3), rendering the split-window into a dual-window
technique. Unfortunately, past measurements at 3.7 pm as taken by the AVHRR instrument
were too noisy for a suitable utilization and the dual-window techniques were less frequently
employed than the split-window method. The ATSR that is flown on the ERS-1 satellite carries
a 3.7 pm channel with low noise amplitudes that allows its full utilization for the atmospheric
correction of nocturnal infrared imagery when deriving sea surface temperature.
In some cases the coefficients being used in (16.3) are tuned to distinct regions in order to optimize the temperature retrieval for specific atmospheric situations (e.g. Llewellyn-Jones et aI.,
1984; Minnett, 1990). This regional tuning is only partly successful since even in small areas
the variability of atmospheric water vapour can reach high values. For example, in the tropical
Pacific Ocean day to day variations of the water-vapour columnn by more than 2 gjcm 2 can
be observed in SSMjI water-vapour measurements. Besides this the regionalization of retrieval
schemes can lead to discontinuities in the derived fields at the boundaries between adjacent
regions so that in large-scale applications temperature gradients can be faked. However, with
respect to studies of limited areas the regionalisation of retrieval schemes might be successful.
A further possibility for the improvement of the surface temperature retrieval is the knowledge
of the actual state of the lower atmosphere. This could be obtained from radiosoundings or
via satellite-derived temperature and moisture profiles. Even more advantageous seems the
direct utilization of additional information from other instruments flying on the same satellites.
Case studies for the synergetic use of different radiometers for the atmospheric correction have
shown improved performances when compared to the use of a two- or three-channel technique
(Schliissel et aI., 1987; Wick et aI., 1992). A disadvantage of this combination of instruments
is the difference of the horizontal resolution of the radiometers in use. While AVHRR resolves
an area of 0.95 km 2 at nadir view an infrared sounder like the High-resolution Infrared Radiation Sounder (HIRS) covers 227 km 2 in a pixel near the sub-satellite track. Computationally
expensive interpolations are necessary in order to match one instrument to the other.
16.2.1 Retrieval techniques
Usually, radiative transfer calculations are used for the development of retrieval techniques.
Each retrieval scheme has to be optimized according to the intrument characteristics because
of different noise behaviour and spectral response functions. It has been shown that the utilization of split-window coefficients developed for one radiometer (AVHRRj2 on board NOAA-7)
in the retrieval of To from another radiometer (AVHRRj2 on board NOAA-9) can introduce
systematic errors up to 0.29 K (Minnett, 1990).
Examples of simulated brightness temperature deficits in the 12 pm channel at different viewing angles of the ATSR are shown in figure 16.2 for 289 globally distributed test cases. The
corrections necessary broadly vary with the surface temperature which is roughly correlated
with the atmospheric water-vapour amount. The slant view through the atmosphere strongly
increases the differences. The values vary between -2 and 16 K at nadir view and up to 22 K