Geoscience Reference
In-Depth Information
Work with Reading University has been undertaken to look at the potential for use of winds
derived from insect returns in fine weather (Rennie et al
.,
2010). This will continue in
collaboration with CAWCR in Australia. Radar returns only give radial winds (i.e. in the
direction of the radar beam) rather than 3-D wind components, so the additional
information in areas of overlapping radars (dual-Doppler) may increase the impact of wind
retrievals in those locations.
5.4.4 Conclusions and further work
1.5 km grid length NWP in the Met Office is showing promise in the very short range
prediction of convection over the UK. Previous sections have highlighted the potential
benefits of using radar derived precipitation rates through latent heat nudging on top of 4D-
Var and of using Doppler radar derived radial winds in 3D-Var.
4D-Var has the potential to exploit higher time frequency observations and to extract more
information from them than 3D-Var. Therefore, research is continuing on the use of high
time frequency Doppler radial winds, direct application of radar derived surface
precipitation rate, and direct and indirect use of multi-elevation volume scan reflectivity in
4D-Var. Although latent heat nudging is still showing benefit in forecasts, it cannot correctly
represent resolved convection where latent heat release occurs in different locations to
surface precipitation, so it is hoped to obtain benefits from direct 4D-Var or indirect 1D-Var
assimilation of the reflectivity data.
Unfortunately, 4D-Var is computationally expensive on the super-computer currently
available to the Met Office. Therefore, research and development is being undertaken with
both 3D-Var and 4D-Var systems. With a super-computer upgrade due in 2012, the aim is to
start running a prototype real-time NWP-based nowcast system in 2012, hopefully with 4D-
Var if the upgrade provides sufficient computer resources.
Due to the tight time constraints imposed by operational schedules, it may be necessary to
move away from use of a time window centred on the analysis time to one finishing at the
analysis time. High quality data sources such as GPS, which provide information on low
level humidity, are currently only available 90 minutes after data time, although less
accurate but more timely data may become available. There are many sources of information
on different variables (e.g. GPS, radar refractivity, satellite imagery and surface observations
for low level humidity). The usefulness of the different data sources will be investigated to
provide an optimum system. The initial experiments reported here were undertaken nested
within the UK 4 km NWP forecast system. Now, the nowcasting system is being tested
embedded in the UK 1.5 km NWP forecast system.
The skill of the convective scale nowcasts is very dependent on the accuracy of the
synoptic forcing conditions both within the nowcast domain itself and the boundary
conditions. Both the UK models and the embedded nowcast system use the same model
and essentially the same data assimilation system. Errors in convective initiation can come
from errors in the synoptic flow either as a result of lack of observations to correct model
errors, or incorrect or sub-optimal use of observations. Finding the best way to extract
synoptic scale and convective scale information from observations in both the nowcast
system itself and in the forcing at the boundaries will be key to improvements in the skill
of the nowcast.
