PERSPECTIVES FROM GODAE
5
et al. (1969) concerning US participation in the then Global Atmospheric
Research Program (GARP): “It is estimated that the data requirements of
computer models are met for only 20 per cent of the earth’s surface. Vast
oceanic regions remain unobserved… the earth-orbiting satellite affords the
opportunity of developing an economically feasible global observing
capability.” Meteorologists were concerned with their ability to observe the
relevant atmospheric variables, at all levels and globally, and to have that
data available each day for models and forecasts. Moreover, on the basis of
progress made with atmospheric models, they wished to test the hypothesis
that models and data assimilation could extend useful predictability and
provide useful forecasts, at lead times several days ahead of what was
possible at that time.
The goals of GARP were effectively (a) deterministic weather
forecasting and (b) understanding climate. The First GARP Global
Experiment (FGGE) was conceived to address the challenges above and set
down several specific goals:
(i)
Development of more realistic models for extended range
forecasting, general circulation studies, and climate.
(ii)
To assess the ultimate limit of predictability of weather systems.
(iii)
To develop more powerful methods for assimilation of
meteorological observations and, in particular, for using nonsynchronous data...
(iv)
To design an optimum composite meteorological observing
system for routine numerical weather prediction.
Bengtsson (1981) discusses the impact of FGGE on numerical weather
prediction, the meteorological counterpart of the systems GODAE is
developing. It is clear that significant progress was made against each of the
goals of FGGE and that that experiment was critical in the development of
modern weather prediction systems. Palmer also showed the evolution of
forecast skill since FGGE, around 2 extra days in lead time in the Northern
Hemisphere, and over 3 for the Southern Hemisphere. This progress has
been made possible by better observations (particularly remote sensing),
better models, faster computers, and most importantly, a vastly improved
knowledge of the dynamics and physics of the atmosphere. The improved
skill however only tells part of the story. The information content of a
modern numerical weather prediction system bears little resemblance to its
predecessors during FGGE. Regional models are often operating at scales of
5-10 km or better, and these broad measures of skill do not capture the
immense value added through finer resolution (indeed, in some cases, the
systems are penalized!). Many forecasts systems are also producing more
than one forecast (ensembles) so that the users can now apply forecasts with
knowledge of the probability of an event occurring. Assimilation systems are
5
et al. (1969) concerning US participation in the then Global Atmospheric
Research Program (GARP): “It is estimated that the data requirements of
computer models are met for only 20 per cent of the earth’s surface. Vast
oceanic regions remain unobserved… the earth-orbiting satellite affords the
opportunity of developing an economically feasible global observing
capability.” Meteorologists were concerned with their ability to observe the
relevant atmospheric variables, at all levels and globally, and to have that
data available each day for models and forecasts. Moreover, on the basis of
progress made with atmospheric models, they wished to test the hypothesis
that models and data assimilation could extend useful predictability and
provide useful forecasts, at lead times several days ahead of what was
possible at that time.
The goals of GARP were effectively (a) deterministic weather
forecasting and (b) understanding climate. The First GARP Global
Experiment (FGGE) was conceived to address the challenges above and set
down several specific goals:
(i)
Development of more realistic models for extended range
forecasting, general circulation studies, and climate.
(ii)
To assess the ultimate limit of predictability of weather systems.
(iii)
To develop more powerful methods for assimilation of
meteorological observations and, in particular, for using nonsynchronous data...
(iv)
To design an optimum composite meteorological observing
system for routine numerical weather prediction.
Bengtsson (1981) discusses the impact of FGGE on numerical weather
prediction, the meteorological counterpart of the systems GODAE is
developing. It is clear that significant progress was made against each of the
goals of FGGE and that that experiment was critical in the development of
modern weather prediction systems. Palmer also showed the evolution of
forecast skill since FGGE, around 2 extra days in lead time in the Northern
Hemisphere, and over 3 for the Southern Hemisphere. This progress has
been made possible by better observations (particularly remote sensing),
better models, faster computers, and most importantly, a vastly improved
knowledge of the dynamics and physics of the atmosphere. The improved
skill however only tells part of the story. The information content of a
modern numerical weather prediction system bears little resemblance to its
predecessors during FGGE. Regional models are often operating at scales of
5-10 km or better, and these broad measures of skill do not capture the
immense value added through finer resolution (indeed, in some cases, the
systems are penalized!). Many forecasts systems are also producing more
than one forecast (ensembles) so that the users can now apply forecasts with
knowledge of the probability of an event occurring. Assimilation systems are