128
e.G. Collier
when implementing operational systems. It is imperative to understand the limitations
of radar, and put in place the necessary processing system. Joss and Lee (1995)
describe a procedure of this type, but other approaches as outlined in Section 63.1
may be adequate. Much will depend upon the operating environment, and the use to
which the data are put.
There continues to be a need to test both radar and satellite measurement techniques. Routine assessment is necessary, both for these radar procedures which use
independent meteorological and mesoscale model data such as the UK Nimrod
System (Harrison et aI., 1995), and for those such as the WPMM which do not use
such data. Whatever techniques are used, it is generally recognised that radar data
should be processed to remove spurious echoes such as ground clutter.
For satellite techniques, as part of the GPCP three (so far) Algorithm Intercomparison Projects (AIPs) have been organised over Japan (1989), over North West Europe
(1991) and over the tropical Pacific Ocean as part of TOGA-CO ARE (1993). These
AlPs should help us to understand a wide range of algorithms and new satellite
instruments. So far they have tended to demonstrate that no one technique outperforms another on all occasions.
6.4.2 The future
Undoubtedly work will continue to improve individual measurement algorithms. It
is necessary to ascertain whether the use of extensive independent meteorological
observations and mesoscale model data are essential to achieve reliable radar estimates of rainfall, or whether the WPMM or some variant is acceptable.
Passive microwave data, particular at frequencies around 85.5 Ghz, seem to offer
improved rainfall estimates from space. Sampling problems will remain, and one
wonders whether it is sensible to keep refining individual algorithms. Perhaps the way
forward is within the context of the data assimilation procedure of a numerical
weather predication model. Numerical models are already capable of reproducing
good rainfall distributions in situations where the rainfall is organised by orography.
Models cannot however, as yet, reproduce convective rainfall, but the assimilation of
satellite and radar data will undoubtedly lead to improvements in the quality of the
model output.
There are dangers, of course, in seeking to improve numerical model rainfall output
in this way. It is all to easy to regard a numerical model rainfall field as "observations". As work on data assimilation proceeds, there is an even greater need for
independent observational datasets, particularly of precipitation which is so variable
in space and time. It is heartening to note that international projects such as BAL TEX
strive to create the most accurate precipitation fields that can be achieved.
In the future there will be developments in both weather radar and satellite instrumentation. Work is already underway to exploit polarisation diversity techniques for
operational ground-based radars, which promises to provide some improvement in
precipitation measurement particularly for high rainfall rates. Undoubtedly the
experience gained in TRMM will lead to improved spacebome active radars, and it
e.G. Collier
when implementing operational systems. It is imperative to understand the limitations
of radar, and put in place the necessary processing system. Joss and Lee (1995)
describe a procedure of this type, but other approaches as outlined in Section 63.1
may be adequate. Much will depend upon the operating environment, and the use to
which the data are put.
There continues to be a need to test both radar and satellite measurement techniques. Routine assessment is necessary, both for these radar procedures which use
independent meteorological and mesoscale model data such as the UK Nimrod
System (Harrison et aI., 1995), and for those such as the WPMM which do not use
such data. Whatever techniques are used, it is generally recognised that radar data
should be processed to remove spurious echoes such as ground clutter.
For satellite techniques, as part of the GPCP three (so far) Algorithm Intercomparison Projects (AIPs) have been organised over Japan (1989), over North West Europe
(1991) and over the tropical Pacific Ocean as part of TOGA-CO ARE (1993). These
AlPs should help us to understand a wide range of algorithms and new satellite
instruments. So far they have tended to demonstrate that no one technique outperforms another on all occasions.
6.4.2 The future
Undoubtedly work will continue to improve individual measurement algorithms. It
is necessary to ascertain whether the use of extensive independent meteorological
observations and mesoscale model data are essential to achieve reliable radar estimates of rainfall, or whether the WPMM or some variant is acceptable.
Passive microwave data, particular at frequencies around 85.5 Ghz, seem to offer
improved rainfall estimates from space. Sampling problems will remain, and one
wonders whether it is sensible to keep refining individual algorithms. Perhaps the way
forward is within the context of the data assimilation procedure of a numerical
weather predication model. Numerical models are already capable of reproducing
good rainfall distributions in situations where the rainfall is organised by orography.
Models cannot however, as yet, reproduce convective rainfall, but the assimilation of
satellite and radar data will undoubtedly lead to improvements in the quality of the
model output.
There are dangers, of course, in seeking to improve numerical model rainfall output
in this way. It is all to easy to regard a numerical model rainfall field as "observations". As work on data assimilation proceeds, there is an even greater need for
independent observational datasets, particularly of precipitation which is so variable
in space and time. It is heartening to note that international projects such as BAL TEX
strive to create the most accurate precipitation fields that can be achieved.
In the future there will be developments in both weather radar and satellite instrumentation. Work is already underway to exploit polarisation diversity techniques for
operational ground-based radars, which promises to provide some improvement in
precipitation measurement particularly for high rainfall rates. Undoubtedly the
experience gained in TRMM will lead to improved spacebome active radars, and it