Geoscience Reference
In-Depth Information
There are still incremental gains to be made by improving the accuracy of the tracking
algorithms and by combining the cell tracking and field advection paradigms into a single
advection scheme. More generally, there is evidence that the cell tracking and field tracking
systems are complementary, so rather than viewing them as competitors there should be
value in developing a way of optimally combining the forecasts from several nowcasting
systems based on an analysis of which system is likely to be providing better nowcasts in
any given situation.
Predicting the initiation and decay of convective storms will continue to be a major focus for
research because gains in this area will lead to significant improvements in the accuracy of
nowcasts beyond 30 minutes. The problem with heuristic and analogue techniques is that
they require large data sets for calibration, and the associated conceptual models that are
developed tend to be location specific. This can be overcome if a way can be found to allow
the algorithms to learn as they go, based on the results of routine real-time verification.
Possibly the major use of radar data in the future will be for assimilation into NWP models
of the national weather services that run radar networks. Empirical advection nowcasting
will continue to provide nowcasts, but for more limited lead times as the NWP models gain
accuracy at shorter lead times. Not all users will be able to afford the costs of a full NWP
system that is able to assimilate radar data and there will continue to be a demand for fast
and cheap rainfall nowcasts for specific purposes.
Forecast errors, rather like death and taxes, will always be with us and the future lies in
using ensembles or other techniques to convey the uncertainty in the current forecast to the
users. Further research on quantifying forecast errors and understanding how they depend
on location and meteorological situation is required before we are able to demonstrate that
the spread in a nowcast ensemble fully represents the uncertainty. There is also a need to
develop probabilistic nowcasting systems that do not only forecast rainfall, but are used to
forecast end-user impacts, for example the traffic capacity of an air-corridor, or the water
level in a river.
8. References
Anagnostou, E. N., Krajewski, W. F. & Smith, J. A. (1999). Uncertainty quantification of
mean-areal radar-rainfall estimates,
Journal of Atmopheric and Oceanic Technol
ogy,
Vol. 16, No. 2, (February 1999), pp. 206-215, Available from
http://dx.doi.org/10.1175/1520-0426(1999)016<0206:UQOMAR>2.0.CO;2
Anderl, B., Attmannspacher, W. & Schultz, G. A. (1976). Accuracy of reservoir inflow
forecasts based on radar rainfall measurements,
Water Resources Res
earch, Vol. 12,
No. 2, pp. 217-223, doi:10.1029/WR012i002p00217
Andersson, T. & Ivarsson, K.-I. (1991). A model for probability nowcasts of accumulated
precipitation using radar.
Journal of Applied Meteorology
, Vol. 30, (January 1991), pp.
135-141
Austin, P. M. (1987). Relation between measured radar reflectivity and surface rainfall.
Monthly Weather Rev
iew, Vol. 115, No. 5, pp. 1053-1070, ISSN 00270644
Austin, G. L. & Bellon, A. (1974). The use of digital weather records for short-term
precipitation forecasting,
Quarterly Journal of the Royal Meteorological Soc
iety, Vol.
100, No. 426, (), pp. 658-664
